Merge with master

.flake8

@@ -15,7 +15,7 @@ per-file-ignores =
     egs/librispeech/ASR/zipformer_mmi/*.py: E501, E203
     egs/librispeech/ASR/zipformer/*.py: E501, E203
     egs/librispeech/ASR/RESULTS.md: E999,
-
+    egs/ljspeech/TTS/vits/*.py: E501, E203
     # invalid escape sequence (cause by tex formular), W605
     icefall/utils.py: E501, W605
 

.github/scripts/aishell/ASR/run.sh

@@ -0,0 +1,343 @@
+#!/usr/bin/env bash
+
+set -ex
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+cd egs/aishell/ASR
+
+function download_test_dev_manifests() {
+  git lfs install
+
+  fbank_url=https://huggingface.co/csukuangfj/aishell-test-dev-manifests
+  log "Downloading pre-commputed fbank from $fbank_url"
+
+  git clone https://huggingface.co/csukuangfj/aishell-test-dev-manifests
+  ln -s $PWD/aishell-test-dev-manifests/data .
+}
+
+function test_transducer_stateless3_2022_06_20() {
+  repo_url=https://huggingface.co/csukuangfj/icefall-aishell-pruned-transducer-stateless3-2022-06-20
+  log "Downloading pre-trained model from $repo_url"
+  git clone $repo_url
+  repo=$(basename $repo_url)
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+  pushd $repo/exp
+  ln -s pretrained-epoch-29-avg-5-torch-1.10.0.pt pretrained.pt
+  popd
+
+  log "test greedy_search with pretrained.py"
+
+  for sym in 1 2 3; do
+    log "Greedy search with --max-sym-per-frame $sym"
+
+    ./pruned_transducer_stateless3/pretrained.py \
+      --method greedy_search \
+      --max-sym-per-frame $sym \
+      --checkpoint $repo/exp/pretrained.pt \
+      --lang-dir $repo/data/lang_char \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+
+  log "test beam search with pretrained.py"
+
+  for method in modified_beam_search beam_search fast_beam_search; do
+    log "$method"
+
+    ./pruned_transducer_stateless3/pretrained.py \
+      --method $method \
+      --beam-size 4 \
+      --checkpoint $repo/exp/pretrained.pt \
+      --lang-dir $repo/data/lang_char \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+
+  echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
+  echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
+  if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
+    mkdir -p pruned_transducer_stateless3/exp
+    ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless3/exp/epoch-999.pt
+    ln -s $PWD/$repo/data/lang_char data/
+
+    ls -lh data
+    ls -lh pruned_transducer_stateless3/exp
+
+    log "Decoding test and dev"
+
+    # use a small value for decoding with CPU
+    max_duration=100
+
+    for method in greedy_search fast_beam_search modified_beam_search; do
+      log "Decoding with $method"
+
+      ./pruned_transducer_stateless3/decode.py \
+        --decoding-method $method \
+        --epoch 999 \
+        --avg 1 \
+        --max-duration $max_duration \
+        --exp-dir pruned_transducer_stateless3/exp
+    done
+
+    rm pruned_transducer_stateless3/exp/*.pt
+  fi
+
+  rm -rf $repo
+}
+
+function test_zipformer_large_2023_10_24() {
+  log "CI testing large model"
+  repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-large-2023-10-24/
+  log "Downloading pre-trained model from $repo_url"
+  git clone $repo_url
+  repo=$(basename $repo_url)
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+  for method in modified_beam_search greedy_search fast_beam_search; do
+    log "$method"
+
+    ./zipformer/pretrained.py \
+      --method $method \
+      --context-size 1 \
+      --checkpoint $repo/exp/pretrained.pt \
+      --tokens $repo/data/lang_char/tokens.txt \
+      --num-encoder-layers 2,2,4,5,4,2 \
+      --feedforward-dim 512,768,1536,2048,1536,768 \
+      --encoder-dim 192,256,512,768,512,256 \
+      --encoder-unmasked-dim 192,192,256,320,256,192 \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+  rm -rf $repo
+}
+
+function test_zipformer_2023_10_24() {
+  repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-2023-10-24/
+  log "Downloading pre-trained model from $repo_url"
+  git clone $repo_url
+  repo=$(basename $repo_url)
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+
+  for method in modified_beam_search greedy_search fast_beam_search; do
+    log "$method"
+
+    ./zipformer/pretrained.py \
+      --method $method \
+      --context-size 1 \
+      --checkpoint $repo/exp/pretrained.pt \
+      --tokens $repo/data/lang_char/tokens.txt \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+  rm -rf $repo
+}
+
+function test_zipformer_small_2023_10_24() {
+  log "CI testing small model"
+  repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-small-2023-10-24/
+  log "Downloading pre-trained model from $repo_url"
+  git clone $repo_url
+  repo=$(basename $repo_url)
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+
+  for method in modified_beam_search greedy_search fast_beam_search; do
+    log "$method"
+
+    ./zipformer/pretrained.py \
+      --method $method \
+      --context-size 1 \
+      --checkpoint $repo/exp/pretrained.pt \
+      --tokens $repo/data/lang_char/tokens.txt \
+      --num-encoder-layers 2,2,2,2,2,2 \
+      --feedforward-dim 512,768,768,768,768,768 \
+      --encoder-dim 192,256,256,256,256,256 \
+      --encoder-unmasked-dim 192,192,192,192,192,192 \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+  rm -rf $repo
+}
+
+function test_transducer_stateless_modified_2022_03_01() {
+  repo_url=https://huggingface.co/csukuangfj/icefall-aishell-transducer-stateless-modified-2022-03-01
+
+  log "Downloading pre-trained model from $repo_url"
+  git lfs install
+  git clone $repo_url
+  repo=$(basename $repo_url)
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+  for sym in 1 2 3; do
+    log "Greedy search with --max-sym-per-frame $sym"
+
+    ./transducer_stateless_modified/pretrained.py \
+      --method greedy_search \
+      --max-sym-per-frame $sym \
+      --checkpoint $repo/exp/pretrained.pt \
+      --lang-dir $repo/data/lang_char \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+
+  for method in modified_beam_search beam_search; do
+    log "$method"
+
+    ./transducer_stateless_modified/pretrained.py \
+      --method $method \
+      --beam-size 4 \
+      --checkpoint $repo/exp/pretrained.pt \
+      --lang-dir $repo/data/lang_char \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+  rm -rf $repo
+}
+
+function test_transducer_stateless_modified_2_2022_03_01() {
+  repo_url=https://huggingface.co/csukuangfj/icefall-aishell-transducer-stateless-modified-2-2022-03-01
+
+  log "Downloading pre-trained model from $repo_url"
+  git lfs install
+  git clone $repo_url
+  repo=$(basename $repo_url)
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+  for sym in 1 2 3; do
+    log "Greedy search with --max-sym-per-frame $sym"
+
+    ./transducer_stateless_modified-2/pretrained.py \
+      --method greedy_search \
+      --max-sym-per-frame $sym \
+      --checkpoint $repo/exp/pretrained.pt \
+      --lang-dir $repo/data/lang_char \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+
+  for method in modified_beam_search beam_search; do
+    log "$method"
+
+    ./transducer_stateless_modified-2/pretrained.py \
+      --method $method \
+      --beam-size 4 \
+      --checkpoint $repo/exp/pretrained.pt \
+      --lang-dir $repo/data/lang_char \
+      $repo/test_wavs/BAC009S0764W0121.wav \
+      $repo/test_wavs/BAC009S0764W0122.wav \
+      $repo/test_wavs/BAC009S0764W0123.wav
+  done
+  rm -rf $repo
+}
+
+function test_conformer_ctc() {
+  repo_url=https://huggingface.co/csukuangfj/icefall_asr_aishell_conformer_ctc
+  log "Downloading pre-trained model from $repo_url"
+  GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+  repo=$(basename $repo_url)
+  pushd $repo
+
+  git lfs pull --include "exp/pretrained.pt"
+  git lfs pull --include "data/lang_char/H.fst"
+  git lfs pull --include "data/lang_char/HL.fst"
+  git lfs pull --include "data/lang_char/HLG.fst"
+
+  popd
+
+  log "Display test files"
+  tree $repo/
+  ls -lh $repo/test_wavs/*.wav
+
+  log "CTC decoding"
+
+  log "Exporting model with torchscript"
+
+  pushd $repo/exp
+  ln -s pretrained.pt epoch-99.pt
+  popd
+
+  ./conformer_ctc/export.py \
+    --epoch 99 \
+    --avg 1 \
+    --exp-dir $repo/exp \
+    --tokens $repo/data/lang_char/tokens.txt \
+    --jit 1
+
+  ls -lh $repo/exp
+
+  ls -lh $repo/data/lang_char
+
+  log "Decoding with H on CPU with OpenFst"
+
+  ./conformer_ctc/jit_pretrained_decode_with_H.py \
+    --nn-model $repo/exp/cpu_jit.pt \
+    --H $repo/data/lang_char/H.fst \
+    --tokens $repo/data/lang_char/tokens.txt \
+    $repo/test_wavs/0.wav \
+    $repo/test_wavs/1.wav \
+    $repo/test_wavs/2.wav
+
+  log "Decoding with HL on CPU with OpenFst"
+
+  ./conformer_ctc/jit_pretrained_decode_with_HL.py \
+    --nn-model $repo/exp/cpu_jit.pt \
+    --HL $repo/data/lang_char/HL.fst \
+    --words $repo/data/lang_char/words.txt \
+    $repo/test_wavs/0.wav \
+    $repo/test_wavs/1.wav \
+    $repo/test_wavs/2.wav
+
+  log "Decoding with HLG on CPU with OpenFst"
+
+  ./conformer_ctc/jit_pretrained_decode_with_HLG.py \
+    --nn-model $repo/exp/cpu_jit.pt \
+    --HLG $repo/data/lang_char/HLG.fst \
+    --words $repo/data/lang_char/words.txt \
+    $repo/test_wavs/0.wav \
+    $repo/test_wavs/1.wav \
+    $repo/test_wavs/2.wav
+
+  rm -rf $repo
+}
+
+download_test_dev_manifests
+test_transducer_stateless3_2022_06_20
+test_zipformer_large_2023_10_24
+test_zipformer_2023_10_24
+test_zipformer_small_2023_10_24
+test_transducer_stateless_modified_2022_03_01
+test_transducer_stateless_modified_2_2022_03_01
+# test_conformer_ctc # fails for torch 1.13.x and torch 2.0.x

.github/scripts/docker/Dockerfile

@@ -0,0 +1,63 @@
+ARG PYTHON_VERSION=3.8
+FROM python:${PYTHON_VERSION}
+
+ARG TORCHAUDIO_VERSION="0.13.0"
+ARG TORCH_VERSION="1.13.0"
+ARG K2_VERSION="1.24.4.dev20231220"
+ARG KALDIFEAT_VERSION="1.25.3.dev20231221"
+
+ARG _K2_VERSION="${K2_VERSION}+cpu.torch${TORCH_VERSION}"
+ARG _KALDIFEAT_VERSION="${KALDIFEAT_VERSION}+cpu.torch${TORCH_VERSION}"
+
+RUN apt-get update -y && \
+    apt-get install -qq -y \
+    ffmpeg \
+    git \
+    git-lfs \
+    graphviz \
+    less \
+    tree \
+    vim \
+    && \
+    apt-get clean && \
+    rm -rf /var/cache/apt/archives /var/lib/apt/lists
+
+
+LABEL authors="Fangjun Kuang <csukuangfj@gmail.com>"
+LABEL k2_version=${_K2_VERSION}
+LABEL kaldifeat_version=${_KALDIFEAT_VERSION}
+LABEL github_repo="https://github.com/k2-fsa/icefall"
+
+# Install dependencies
+RUN pip install --no-cache-dir \
+      torch==${TORCH_VERSION} torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/cpu/torch_stable.html \
+      k2==${_K2_VERSION} -f https://k2-fsa.github.io/k2/cpu.html \
+      \
+      git+https://github.com/lhotse-speech/lhotse \
+      kaldifeat==${_KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cpu.html \
+      dill \
+      graphviz \
+      kaldi-decoder \
+      kaldi_native_io \
+      kaldialign \
+      kaldifst \
+      kaldilm \
+      matplotlib \
+      multi_quantization \
+      numpy \
+      onnx \
+      onnxmltools \
+      onnxruntime \
+      pytest \
+      sentencepiece>=0.1.96 \
+      six \
+      tensorboard \
+      typeguard
+
+# RUN git clone https://github.com/k2-fsa/icefall /workspace/icefall && \
+#     cd /workspace/icefall && \
+#     pip install --no-cache-dir -r requirements.txt
+#
+# ENV PYTHONPATH /workspace/icefall:$PYTHONPATH
+#
+# WORKDIR /workspace/icefall

.github/scripts/docker/generate_build_matrix.py

@@ -0,0 +1,79 @@
+#!/usr/bin/env python3
+# Copyright    2023  Xiaomi Corp.        (authors: Fangjun Kuang)
+
+
+import json
+
+
+def version_gt(a, b):
+    a_major, a_minor = a.split(".")[:2]
+    b_major, b_minor = b.split(".")[:2]
+    if a_major > b_major:
+        return True
+
+    if a_major == b_major and a_minor > b_minor:
+        return True
+
+    return False
+
+
+def version_ge(a, b):
+    a_major, a_minor = a.split(".")[:2]
+    b_major, b_minor = b.split(".")[:2]
+    if a_major > b_major:
+        return True
+
+    if a_major == b_major and a_minor >= b_minor:
+        return True
+
+    return False
+
+
+def get_torchaudio_version(torch_version):
+    if torch_version == "1.13.0":
+        return "0.13.0"
+    elif torch_version == "1.13.1":
+        return "0.13.1"
+    elif torch_version == "2.0.0":
+        return "2.0.1"
+    elif torch_version == "2.0.1":
+        return "2.0.2"
+    else:
+        return torch_version
+
+
+def get_matrix():
+    k2_version = "1.24.4.dev20231220"
+    kaldifeat_version = "1.25.3.dev20231221"
+    version = "1.2"
+    python_version = ["3.8", "3.9", "3.10", "3.11"]
+    torch_version = ["1.13.0", "1.13.1", "2.0.0", "2.0.1", "2.1.0", "2.1.1", "2.1.2"]
+
+    matrix = []
+    for p in python_version:
+        for t in torch_version:
+            # torchaudio <= 1.13.x supports only python <= 3.10
+
+            if version_gt(p, "3.10") and not version_gt(t, "2.0"):
+                continue
+
+            matrix.append(
+                {
+                    "k2-version": k2_version,
+                    "kaldifeat-version": kaldifeat_version,
+                    "version": version,
+                    "python-version": p,
+                    "torch-version": t,
+                    "torchaudio-version": get_torchaudio_version(t),
+                }
+            )
+    return matrix
+
+
+def main():
+    matrix = get_matrix()
+    print(json.dumps({"include": matrix}))
+
+
+if __name__ == "__main__":
+    main()

.github/scripts/multi-zh-hans.sh

@@ -0,0 +1,158 @@
+#!/usr/bin/env bash
+
+set -ex
+
+git config --global user.name "k2-fsa"
+git config --global user.email "csukuangfj@gmail.com"
+git config --global lfs.allowincompletepush true
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+log "pwd: $PWD"
+
+cd egs/multi_zh-hans/ASR
+
+repo_url=https://huggingface.co/zrjin/icefall-asr-multi-zh-hans-zipformer-ctc-streaming-2023-11-05
+log "Downloading pre-trained model from $repo_url"
+GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+repo=$(basename $repo_url)
+
+pushd $repo
+cd exp/
+git lfs pull --include pretrained.pt
+rm -fv epoch-20.pt
+rm -fv *.onnx
+ln -s pretrained.pt epoch-20.pt
+cd ../data/lang_bpe_2000
+ls -lh
+git lfs pull --include L.pt L_disambig.pt Linv.pt bpe.model
+git lfs pull --include "*.model"
+ls -lh
+popd
+
+log "----------------------------------------"
+log "Export streaming ONNX CTC models "
+log "----------------------------------------"
+./zipformer/export-onnx-streaming-ctc.py \
+  --exp-dir $repo/exp \
+  --tokens $repo/data/lang_bpe_2000/tokens.txt \
+  --causal 1 \
+  --avg 1 \
+  --epoch 20 \
+  --use-averaged-model 0 \
+  --chunk-size 16 \
+  --left-context-frames 128 \
+  --use-ctc 1
+
+ls -lh $repo/exp/
+
+log "------------------------------------------------------------"
+log "Test exported streaming ONNX CTC models (greedy search)     "
+log "------------------------------------------------------------"
+
+test_wavs=(
+DEV_T0000000000.wav
+DEV_T0000000001.wav
+DEV_T0000000002.wav
+TEST_MEETING_T0000000113.wav
+TEST_MEETING_T0000000219.wav
+TEST_MEETING_T0000000351.wav
+)
+
+for w in ${test_wavs[@]}; do
+  ./zipformer/onnx_pretrained-streaming-ctc.py \
+    --model-filename $repo/exp/ctc-epoch-20-avg-1-chunk-16-left-128.int8.onnx \
+    --tokens $repo/data/lang_bpe_2000/tokens.txt \
+    $repo/test_wavs/$w
+done
+
+log "Upload onnx CTC models to huggingface"
+url=https://huggingface.co/k2-fsa/sherpa-onnx-streaming-zipformer-ctc-multi-zh-hans-2023-12-13
+GIT_LFS_SKIP_SMUDGE=1 git clone $url
+dst=$(basename $url)
+cp -v $repo/exp/ctc*.onnx $dst
+cp -v $repo/data/lang_bpe_2000/tokens.txt $dst
+cp -v $repo/data/lang_bpe_2000/bpe.model $dst
+mkdir -p $dst/test_wavs
+cp -v $repo/test_wavs/*.wav $dst/test_wavs
+cd $dst
+git lfs track "*.onnx" "bpe.model"
+ls -lh
+file bpe.model
+git status
+git add .
+git commit -m "upload model" && git push https://k2-fsa:${HF_TOKEN}@huggingface.co/k2-fsa/$dst main || true
+
+log "Upload models to https://github.com/k2-fsa/sherpa-onnx"
+rm -rf .git
+rm -fv .gitattributes
+cd ..
+tar cjfv $dst.tar.bz2 $dst
+ls -lh *.tar.bz2
+mv -v $dst.tar.bz2 ../../../
+
+log "----------------------------------------"
+log "Export streaming ONNX transducer models "
+log "----------------------------------------"
+
+./zipformer/export-onnx-streaming.py \
+  --exp-dir $repo/exp \
+  --tokens $repo/data/lang_bpe_2000/tokens.txt \
+  --causal 1 \
+  --avg 1 \
+  --epoch 20 \
+  --use-averaged-model 0 \
+  --chunk-size 16 \
+  --left-context-frames 128 \
+  --use-ctc 0
+
+ls -lh $repo/exp
+
+log "------------------------------------------------------------"
+log "Test exported streaming ONNX transducer models (Python code)"
+log "------------------------------------------------------------"
+
+log "test fp32"
+./zipformer/onnx_pretrained-streaming.py \
+  --encoder-model-filename $repo/exp/encoder-epoch-20-avg-1-chunk-16-left-128.onnx \
+  --decoder-model-filename $repo/exp/decoder-epoch-20-avg-1-chunk-16-left-128.onnx \
+  --joiner-model-filename $repo/exp/joiner-epoch-20-avg-1-chunk-16-left-128.onnx \
+  --tokens $repo/data/lang_bpe_2000/tokens.txt \
+  $repo/test_wavs/DEV_T0000000000.wav
+
+log "test int8"
+./zipformer/onnx_pretrained-streaming.py \
+  --encoder-model-filename $repo/exp/encoder-epoch-20-avg-1-chunk-16-left-128.int8.onnx \
+  --decoder-model-filename $repo/exp/decoder-epoch-20-avg-1-chunk-16-left-128.onnx \
+  --joiner-model-filename $repo/exp/joiner-epoch-20-avg-1-chunk-16-left-128.int8.onnx \
+  --tokens $repo/data/lang_bpe_2000/tokens.txt \
+  $repo/test_wavs/DEV_T0000000000.wav
+
+log "Upload onnx transducer models to huggingface"
+
+url=https://huggingface.co/k2-fsa/sherpa-onnx-streaming-zipformer-multi-zh-hans-2023-12-12
+GIT_LFS_SKIP_SMUDGE=1 git clone $url
+dst=$(basename $url)
+cp -v $repo/exp/encoder*.onnx $dst
+cp -v $repo/exp/decoder*.onnx $dst
+cp -v $repo/exp/joiner*.onnx $dst
+cp -v $repo/data/lang_bpe_2000/tokens.txt $dst
+cp -v $repo/data/lang_bpe_2000/bpe.model $dst
+mkdir -p $dst/test_wavs
+cp -v $repo/test_wavs/*.wav $dst/test_wavs
+cd $dst
+git lfs track "*.onnx" bpe.model
+git add .
+git commit -m "upload model" && git push https://k2-fsa:${HF_TOKEN}@huggingface.co/k2-fsa/$dst main || true
+
+log "Upload models to https://github.com/k2-fsa/sherpa-onnx"
+rm -rf .git
+rm -fv .gitattributes
+cd ..
+tar cjfv $dst.tar.bz2 $dst
+ls -lh *.tar.bz2
+mv -v $dst.tar.bz2 ../../../

.github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh

@@ -1,87 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/aishell/ASR
-
-git lfs install
-
-fbank_url=https://huggingface.co/csukuangfj/aishell-test-dev-manifests
-log "Downloading pre-commputed fbank from $fbank_url"
-
-git clone https://huggingface.co/csukuangfj/aishell-test-dev-manifests
-ln -s $PWD/aishell-test-dev-manifests/data .
-
-repo_url=https://huggingface.co/csukuangfj/icefall-aishell-pruned-transducer-stateless3-2022-06-20
-log "Downloading pre-trained model from $repo_url"
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-ln -s pretrained-epoch-29-avg-5-torch-1.10.0.pt pretrained.pt
-popd
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless3/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_char \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless3/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_char \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless3/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless3/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_char data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless3/exp
-
-  log "Decoding test and dev"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless3/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless3/exp
-  done
-
-  rm pruned_transducer_stateless3/exp/*.pt
-fi

.github/scripts/run-aishell-zipformer-2023-10-24.sh

@@ -1,103 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/aishell/ASR
-
-git lfs install
-
-fbank_url=https://huggingface.co/csukuangfj/aishell-test-dev-manifests
-log "Downloading pre-commputed fbank from $fbank_url"
-
-git clone https://huggingface.co/csukuangfj/aishell-test-dev-manifests
-ln -s $PWD/aishell-test-dev-manifests/data .
-
-log "======================="
-log "CI testing large model"
-repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-large-2023-10-24/
-log "Downloading pre-trained model from $repo_url"
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for method in modified_beam_search greedy_search fast_beam_search; do
-  log "$method"
-
-  ./zipformer/pretrained.py \
-    --method $method \
-    --context-size 1 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_char/tokens.txt \
-    --num-encoder-layers 2,2,4,5,4,2 \
-    --feedforward-dim 512,768,1536,2048,1536,768 \
-    --encoder-dim 192,256,512,768,512,256 \
-    --encoder-unmasked-dim 192,192,256,320,256,192 \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-
-log "======================="
-log "CI testing medium model"
-repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-2023-10-24/
-log "Downloading pre-trained model from $repo_url"
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-
-for method in modified_beam_search greedy_search fast_beam_search; do
-  log "$method"
-
-  ./zipformer/pretrained.py \
-    --method $method \
-    --context-size 1 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_char/tokens.txt \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-
-
-log "======================="
-log "CI testing small model"
-repo_url=https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-small-2023-10-24/
-log "Downloading pre-trained model from $repo_url"
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-
-for method in modified_beam_search greedy_search fast_beam_search; do
-  log "$method"
-
-  ./zipformer/pretrained.py \
-    --method $method \
-    --context-size 1 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_char/tokens.txt \
-    --num-encoder-layers 2,2,2,2,2,2 \
-    --feedforward-dim 512,768,768,768,768,768 \
-    --encoder-dim 192,256,256,256,256,256 \
-    --encoder-unmasked-dim 192,192,192,192,192,192 \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-

.github/scripts/run-gigaspeech-zipformer-2023-10-17.sh

@@ -26,16 +26,80 @@ git lfs pull --include "data/lang_bpe_500/bpe.model"
 git lfs pull --include "data/lang_bpe_500/tokens.txt"
 git lfs pull --include "exp/jit_script.pt"
 git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
+rm epoch-30.pt
-ls -lh *.pt
+ln -s pretrained.pt epoch-30.pt
+rm *.onnx
+ls -lh
 popd
 
+log "----------------------------------------"
+log "Export ONNX transducer models "
+log "----------------------------------------"
+
+./zipformer/export-onnx.py \
+  --tokens $repo/data/lang_bpe_500/tokens.txt \
+  --use-averaged-model 0 \
+  --epoch 30 \
+  --avg 1 \
+  --exp-dir $repo/exp
+
+ls -lh $repo/exp
+
+log "------------------------------------------------------------"
+log "Test exported ONNX transducer models (Python code)          "
+log "------------------------------------------------------------"
+
+log "test fp32"
+./zipformer/onnx_pretrained.py \
+  --encoder-model-filename $repo/exp/encoder-epoch-30-avg-1.onnx \
+  --decoder-model-filename $repo/exp/decoder-epoch-30-avg-1.onnx \
+  --joiner-model-filename $repo/exp/joiner-epoch-30-avg-1.onnx \
+  --tokens $repo/data/lang_bpe_500/tokens.txt \
+  $repo/test_wavs/1089-134686-0001.wav \
+  $repo/test_wavs/1221-135766-0001.wav \
+  $repo/test_wavs/1221-135766-0002.wav
+
+log "test int8"
+./zipformer/onnx_pretrained.py \
+  --encoder-model-filename $repo/exp/encoder-epoch-30-avg-1.int8.onnx \
+  --decoder-model-filename $repo/exp/decoder-epoch-30-avg-1.onnx \
+  --joiner-model-filename $repo/exp/joiner-epoch-30-avg-1.int8.onnx \
+  --tokens $repo/data/lang_bpe_500/tokens.txt \
+  $repo/test_wavs/1089-134686-0001.wav \
+  $repo/test_wavs/1221-135766-0001.wav \
+  $repo/test_wavs/1221-135766-0002.wav
+
+log "Upload models to huggingface"
+git config --global user.name "k2-fsa"
+git config --global user.email "xxx@gmail.com"
+
+url=https://huggingface.co/k2-fsa/sherpa-onnx-zipformer-gigaspeech-2023-12-12
+GIT_LFS_SKIP_SMUDGE=1 git clone $url
+dst=$(basename $url)
+cp -v $repo/exp/*.onnx $dst
+cp -v $repo/data/lang_bpe_500/tokens.txt $dst
+cp -v $repo/data/lang_bpe_500/bpe.model $dst
+mkdir -p $dst/test_wavs
+cp -v $repo/test_wavs/*.wav $dst/test_wavs
+cd $dst
+git lfs track "*.onnx"
+git add .
+git commit -m "upload model" && git push https://k2-fsa:${HF_TOKEN}@huggingface.co/k2-fsa/$dst main || true
+
+log "Upload models to https://github.com/k2-fsa/sherpa-onnx"
+rm -rf .git
+rm -fv .gitattributes
+cd ..
+tar cjfv $dst.tar.bz2 $dst
+ls -lh
+mv -v $dst.tar.bz2 ../../../
+
 log "Export to torchscript model"
 ./zipformer/export.py \
   --exp-dir $repo/exp \
   --use-averaged-model false \
   --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
+  --epoch 30 \
   --avg 1 \
   --jit 1
 
@@ -67,7 +131,7 @@ echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
 echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
 if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
   mkdir -p zipformer/exp
-  ln -s $PWD/$repo/exp/pretrained.pt zipformer/exp/epoch-999.pt
+  ln -s $PWD/$repo/exp/pretrained.pt zipformer/exp/epoch-30.pt
   ln -s $PWD/$repo/data/lang_bpe_500 data/
 
   ls -lh data
@@ -83,7 +147,7 @@ if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" ==
 
     ./zipformer/decode.py \
       --decoding-method $method \
-      --epoch 999 \
+      --epoch 30 \
       --avg 1 \
       --use-averaged-model 0 \
       --max-duration $max_duration \

.github/scripts/run-librispeech-conformer-ctc3-2022-11-28.sh

@@ -1,122 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-conformer-ctc3-2022-11-27
-
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/HLG.pt"
-git lfs pull --include "data/lang_bpe_500/L.pt"
-git lfs pull --include "data/lang_bpe_500/LG.pt"
-git lfs pull --include "data/lang_bpe_500/Linv.pt"
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "data/lm/G_4_gram.pt"
-git lfs pull --include "exp/jit_trace.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Decode with models exported by torch.jit.trace()"
-
-for m in ctc-decoding 1best; do
-  ./conformer_ctc3/jit_pretrained.py \
-    --model-filename $repo/exp/jit_trace.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --bpe-model $repo/data/lang_bpe_500/bpe.model \
-    --G $repo/data/lm/G_4_gram.pt \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-log "Export to torchscript model"
-
-./conformer_ctc3/export.py \
-  --exp-dir $repo/exp \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --jit-trace 1 \
-  --epoch 99 \
-  --avg 1 \
-  --use-averaged-model 0
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.trace()"
-
-for m in ctc-decoding 1best; do
-  ./conformer_ctc3/jit_pretrained.py \
-    --model-filename $repo/exp/jit_trace.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --bpe-model $repo/data/lang_bpe_500/bpe.model \
-    --G $repo/data/lm/G_4_gram.pt \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for m in ctc-decoding 1best; do
-  ./conformer_ctc3/pretrained.py \
-    --checkpoint $repo/exp/pretrained.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --G $repo/data/lm/G_4_gram.pt \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode" ]]; then
-  mkdir -p conformer_ctc3/exp
-  ln -s $PWD/$repo/exp/pretrained.pt conformer_ctc3/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh conformer_ctc3/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in ctc-decoding 1best; do
-    log "Decoding with $method"
-    ./conformer_ctc3/decode.py \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --exp-dir conformer_ctc3/exp/ \
-      --max-duration $max_duration \
-      --decoding-method $method \
-      --lm-dir data/lm
-  done
-
-  rm conformer_ctc3/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless-2022-03-12.sh

@@ -1,77 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless-2022-03-12
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in fast_beam_search modified_beam_search beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless/exp
-  done
-
-  rm pruned_transducer_stateless/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless2-2022-04-29.sh

@@ -1,86 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless2-2022-04-29
-
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-pushd $repo
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/pretrained-epoch-38-avg-10.pt"
-popd
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-ln -s pretrained-epoch-38-avg-10.pt pretrained.pt
-popd
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless2/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless2/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless2/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless2/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless2/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless2/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless2/exp
-  done
-
-  rm pruned_transducer_stateless2/exp/*.pt
-  rm -r data/lang_bpe_500
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-04-29.sh

@@ -1,85 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-04-29
-
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-pushd $repo
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/pretrained-epoch-25-avg-6.pt"
-popd
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-ln -s pretrained-epoch-25-avg-6.pt pretrained.pt
-popd
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless3/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless3/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless3/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless3/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless3/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless3/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless3/exp
-  done
-
-  rm pruned_transducer_stateless3/exp/*.pt
-  rm -r data/lang_bpe_500
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-05-13.sh

@@ -1,123 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless3-2022-05-13
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-ln -s pretrained-iter-1224000-avg-14.pt pretrained.pt
-ln -s pretrained-iter-1224000-avg-14.pt epoch-99.pt
-popd
-
-
-log "Export to torchscript model"
-./pruned_transducer_stateless3/export.py \
-  --exp-dir $repo/exp \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-./pruned_transducer_stateless3/export.py \
-  --exp-dir $repo/exp \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit-trace 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.trace()"
-
-./pruned_transducer_stateless3/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --encoder-model-filename $repo/exp/encoder_jit_trace.pt \
-  --decoder-model-filename $repo/exp/decoder_jit_trace.pt \
-  --joiner-model-filename $repo/exp/joiner_jit_trace.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless3/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --encoder-model-filename $repo/exp/encoder_jit_script.pt \
-  --decoder-model-filename $repo/exp/decoder_jit_script.pt \
-  --joiner-model-filename $repo/exp/joiner_jit_script.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless3/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless3/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless3/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless3/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless3/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless3/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless3/exp
-  done
-
-  rm pruned_transducer_stateless3/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless5-2022-05-13.sh

@@ -1,100 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless5-2022-05-13
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-ln -s pretrained-epoch-39-avg-7.pt pretrained.pt
-popd
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless5/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --num-encoder-layers 18 \
-    --dim-feedforward 2048 \
-    --nhead 8 \
-    --encoder-dim 512 \
-    --decoder-dim 512 \
-    --joiner-dim 512 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless5/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav \
-    --num-encoder-layers 18 \
-    --dim-feedforward 2048 \
-    --nhead 8 \
-    --encoder-dim 512 \
-    --decoder-dim 512 \
-    --joiner-dim 512
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless5/exp
-  ln -s $PWD/$repo/exp/pretrained-epoch-39-avg-7.pt pruned_transducer_stateless5/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless5/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless5/decode.py \
-      --decoding-method $method \
-      --use-averaged-model 0 \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless5/exp \
-      --num-encoder-layers 18 \
-      --dim-feedforward 2048 \
-      --nhead 8 \
-      --encoder-dim 512 \
-      --decoder-dim 512 \
-      --joiner-dim 512
-  done
-
-  rm pruned_transducer_stateless5/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless7-2022-11-11.sh

@@ -1,106 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless7-2022-11-11
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/cpu_jit.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./pruned_transducer_stateless7/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless7/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless7/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless7/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless7/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless7/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless7/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless7/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless7/exp
-  done
-
-  rm pruned_transducer_stateless7/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-2022-12-01.sh

@@ -1,150 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-pruned-transducer-stateless7-ctc-2022-12-01
-
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/HLG.pt"
-git lfs pull --include "data/lang_bpe_500/L.pt"
-git lfs pull --include "data/lang_bpe_500/LG.pt"
-git lfs pull --include "data/lang_bpe_500/Linv.pt"
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "data/lm/G_4_gram.pt"
-git lfs pull --include "exp/cpu_jit.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./pruned_transducer_stateless7_ctc/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless7_ctc/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-for m in ctc-decoding 1best; do
-  ./pruned_transducer_stateless7_ctc/jit_pretrained_ctc.py \
-    --model-filename $repo/exp/cpu_jit.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --bpe-model $repo/data/lang_bpe_500/bpe.model \
-    --G $repo/data/lm/G_4_gram.pt \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless7_ctc/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless7_ctc/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for m in ctc-decoding 1best; do
-  ./pruned_transducer_stateless7_ctc/pretrained_ctc.py \
-    --checkpoint $repo/exp/pretrained.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --bpe-model $repo/data/lang_bpe_500/bpe.model \
-    --G $repo/data/lm/G_4_gram.pt \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless7_ctc/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless7_ctc/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless7_ctc/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless7_ctc/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless7_ctc/exp
-  done
-
-  for m in ctc-decoding 1best; do
-    ./pruned_transducer_stateless7_ctc/ctc_decode.py \
-        --epoch 999 \
-        --avg 1 \
-        --exp-dir ./pruned_transducer_stateless7_ctc/exp \
-        --max-duration $max_duration \
-        --use-averaged-model 0 \
-        --decoding-method $m \
-        --hlg-scale 0.6 \
-        --lm-dir data/lm
-  done
-
-  rm pruned_transducer_stateless7_ctc/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-bs-2023-01-29.sh

@@ -1,147 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/yfyeung/icefall-asr-librispeech-pruned_transducer_stateless7_ctc_bs-2023-01-29
-
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/L.pt"
-git lfs pull --include "data/lang_bpe_500/LG.pt"
-git lfs pull --include "data/lang_bpe_500/HLG.pt"
-git lfs pull --include "data/lang_bpe_500/Linv.pt"
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/cpu_jit.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./pruned_transducer_stateless7_ctc_bs/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless7_ctc_bs/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-for m in ctc-decoding 1best; do
-  ./pruned_transducer_stateless7_ctc_bs/jit_pretrained_ctc.py \
-    --model-filename $repo/exp/cpu_jit.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --bpe-model $repo/data/lang_bpe_500/bpe.model \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless7_ctc_bs/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless7_ctc_bs/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for m in ctc-decoding 1best; do
-  ./pruned_transducer_stateless7_ctc_bs/pretrained_ctc.py \
-    --checkpoint $repo/exp/pretrained.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --bpe-model $repo/data/lang_bpe_500/bpe.model \
-    --method $m \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless7_ctc_bs/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless7_ctc_bs/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless7_ctc_bs/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless7_ctc_bs/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless7_ctc_bs/exp
-  done
-
-  for m in ctc-decoding 1best; do
-    ./pruned_transducer_stateless7_ctc_bs/ctc_decode.py \
-        --epoch 999 \
-        --avg 1 \
-        --exp-dir ./pruned_transducer_stateless7_ctc_bs/exp \
-        --max-duration $max_duration \
-        --use-averaged-model 0 \
-        --decoding-method $m \
-        --hlg-scale 0.6
-  done
-  
-  rm pruned_transducer_stateless7_ctc_bs/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless7-streaming-2022-12-29.sh

@@ -1,148 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-pruned-transducer-stateless7-streaming-2022-12-29
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/cpu_jit.pt"
-git lfs pull --include "exp/pretrained.pt"
-git lfs pull --include "exp/encoder_jit_trace.pt"
-git lfs pull --include "exp/decoder_jit_trace.pt"
-git lfs pull --include "exp/joiner_jit_trace.pt"
-cd exp
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./pruned_transducer_stateless7_streaming/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --decode-chunk-len 32 \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless7_streaming/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  --decode-chunk-len 32 \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "Export to torchscript model by torch.jit.trace()"
-./pruned_transducer_stateless7_streaming/jit_trace_export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --decode-chunk-len 32 \
-  --epoch 99 \
-  --avg 1
-
-log "Decode with models exported by torch.jit.trace()"
-
-./pruned_transducer_stateless7_streaming/jit_trace_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --encoder-model-filename $repo/exp/encoder_jit_trace.pt \
-  --decoder-model-filename $repo/exp/decoder_jit_trace.pt \
-  --joiner-model-filename $repo/exp/joiner_jit_trace.pt \
-  --decode-chunk-len 32 \
-  $repo/test_wavs/1089-134686-0001.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless7_streaming/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --decode-chunk-len 32 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless7_streaming/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --decode-chunk-len 32 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless7_streaming/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless7_streaming/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless7_streaming/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-  num_decode_stream=200
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "decoding with $method"
-
-    ./pruned_transducer_stateless7_streaming/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --decode-chunk-len 32 \
-      --exp-dir pruned_transducer_stateless7_streaming/exp
-  done
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless7_streaming/streaming_decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --decode-chunk-len 32 \
-      --num-decode-streams $num_decode_stream
-      --exp-dir pruned_transducer_stateless7_streaming/exp
-  done
-
-  rm pruned_transducer_stateless7_streaming/exp/*.pt
-fi

.github/scripts/run-librispeech-pruned-transducer-stateless8-2022-11-14.sh

@@ -1,115 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-pruned-transducer-stateless8-2022-11-14
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/cpu_jit.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless8/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "Export to torchscript model"
-./pruned_transducer_stateless8/export.py \
-  --exp-dir $repo/exp \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --use-averaged-model false \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./pruned_transducer_stateless8/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless8/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless8/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless8/exp
-  ln -s $PWD/$repo/exp/pretrained.pt pruned_transducer_stateless8/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless8/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./pruned_transducer_stateless8/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless8/exp
-  done
-
-  rm pruned_transducer_stateless8/exp/*.pt
-fi

.github/scripts/run-librispeech-streaming-pruned-transducer-stateless2-2022-06-26.sh

@@ -1,101 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/pkufool/icefall_librispeech_streaming_pruned_transducer_stateless2_20220625
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-ln -s pretrained-epoch-24-avg-10.pt pretrained.pt
-popd
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./pruned_transducer_stateless2/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --simulate-streaming 1 \
-    --causal-convolution 1 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./pruned_transducer_stateless2/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --simulate-streaming 1 \
-    --causal-convolution 1 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p pruned_transducer_stateless2/exp
-  ln -s $PWD/$repo/exp/pretrained-epoch-24-avg-10.pt pruned_transducer_stateless2/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh pruned_transducer_stateless2/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Simulate streaming decoding with $method"
-
-    ./pruned_transducer_stateless2/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir pruned_transducer_stateless2/exp \
-      --simulate-streaming 1 \
-      --causal-convolution 1
-  done
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Real streaming decoding with $method"
-
-    ./pruned_transducer_stateless2/streaming_decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --num-decode-streams 100 \
-      --exp-dir pruned_transducer_stateless2/exp \
-      --left-context 32 \
-      --decode-chunk-size 8 \
-      --right-context 0
-  done
-
-  rm pruned_transducer_stateless2/exp/*.pt
-fi

.github/scripts/run-librispeech-streaming-zipformer-2023-05-18.sh

@@ -1,116 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-streaming-zipformer-2023-05-17
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "data/lang_bpe_500/tokens.txt"
-git lfs pull --include "exp/jit_script_chunk_16_left_128.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./zipformer/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --causal 1 \
-  --chunk-size 16 \
-  --left-context-frames 128 \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./zipformer/jit_pretrained_streaming.py \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --nn-model-filename $repo/exp/jit_script_chunk_16_left_128.pt \
-  $repo/test_wavs/1089-134686-0001.wav
-
-for method in greedy_search modified_beam_search fast_beam_search; do
-  log "$method"
-
-  ./zipformer/pretrained.py \
-    --causal 1 \
-    --chunk-size 16 \
-    --left-context-frames 128 \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p zipformer/exp
-  ln -s $PWD/$repo/exp/pretrained.pt zipformer/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh zipformer/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Simulated streaming decoding with $method"
-
-    ./zipformer/decode.py \
-      --causal 1 \
-      --chunk-size 16 \
-      --left-context-frames 128 \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir zipformer/exp
-  done
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Chunk-wise streaming decoding with $method"
-
-    ./zipformer/streaming_decode.py \
-      --causal 1 \
-      --chunk-size 16 \
-      --left-context-frames 128 \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir zipformer/exp
-  done
-
-  rm zipformer/exp/*.pt
-fi

.github/scripts/run-librispeech-transducer-stateless2-2022-04-19.sh

@@ -1,77 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-transducer-stateless2-torchaudio-2022-04-19
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./transducer_stateless2/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in fast_beam_search modified_beam_search beam_search; do
-  log "$method"
-
-  ./transducer_stateless2/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p transducer_stateless2/exp
-  ln -s $PWD/$repo/exp/pretrained.pt transducer_stateless2/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh transducer_stateless2/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./transducer_stateless2/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir transducer_stateless2/exp
-  done
-
-  rm transducer_stateless2/exp/*.pt
-fi

.github/scripts/run-librispeech-zipformer-2023-05-18.sh

@@ -1,94 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-zipformer-2023-05-15
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "data/lang_bpe_500/tokens.txt"
-git lfs pull --include "exp/jit_script.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./zipformer/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./zipformer/jit_pretrained.py \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --nn-model-filename $repo/exp/jit_script.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-for method in greedy_search modified_beam_search fast_beam_search; do
-  log "$method"
-
-  ./zipformer/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p zipformer/exp
-  ln -s $PWD/$repo/exp/pretrained.pt zipformer/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh zipformer/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./zipformer/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir zipformer/exp
-  done
-
-  rm zipformer/exp/*.pt
-fi

.github/scripts/run-librispeech-zipformer-ctc-2023-06-14.sh

@@ -1,117 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-zipformer-transducer-ctc-2023-06-13
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "data/lang_bpe_500/tokens.txt"
-git lfs pull --include "data/lang_bpe_500/HLG.pt"
-git lfs pull --include "data/lang_bpe_500/L.pt"
-git lfs pull --include "data/lang_bpe_500/LG.pt"
-git lfs pull --include "data/lang_bpe_500/Linv.pt"
-git lfs pull --include "data/lm/G_4_gram.pt"
-git lfs pull --include "exp/jit_script.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./zipformer/export.py \
-  --exp-dir $repo/exp \
-  --use-transducer 1 \
-  --use-ctc 1 \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-for method in ctc-decoding 1best; do
-  ./zipformer/jit_pretrained_ctc.py \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --model-filename $repo/exp/jit_script.pt \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --G $repo/data/lm/G_4_gram.pt \
-    --method $method \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in ctc-decoding 1best; do
-  log "$method"
-
-  ./zipformer/pretrained_ctc.py \
-    --use-transducer 1 \
-    --use-ctc 1 \
-    --method $method \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    --HLG $repo/data/lang_bpe_500/HLG.pt \
-    --G $repo/data/lm/G_4_gram.pt \
-    --words-file $repo/data/lang_bpe_500/words.txt  \
-    --sample-rate 16000 \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p zipformer/exp
-  ln -s $PWD/$repo/exp/pretrained.pt zipformer/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh zipformer/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in ctc-decoding 1best; do
-    log "Decoding with $method"
-
-    ./zipformer/ctc_decode.py \
-      --use-transducer 1 \
-      --use-ctc 1 \
-      --decoding-method $method \
-      --nbest-scale 1.0 \
-      --hlg-scale 0.6 \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --max-duration $max_duration \
-      --exp-dir zipformer/exp
-  done
-
-  rm zipformer/exp/*.pt
-fi

.github/scripts/run-librispeech-zipformer-mmi-2022-12-08.sh

@@ -1,102 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/Zengwei/icefall-asr-librispeech-zipformer-mmi-2022-12-08
-
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-pushd $repo/exp
-git lfs pull --include "data/lang_bpe_500/3gram.pt"
-git lfs pull --include "data/lang_bpe_500/4gram.pt"
-git lfs pull --include "data/lang_bpe_500/L.pt"
-git lfs pull --include "data/lang_bpe_500/LG.pt"
-git lfs pull --include "data/lang_bpe_500/Linv.pt"
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "exp/cpu_jit.pt"
-git lfs pull --include "exp/pretrained.pt"
-ln -s pretrained.pt epoch-99.pt
-ls -lh *.pt
-popd
-
-log "Export to torchscript model"
-./zipformer_mmi/export.py \
-  --exp-dir $repo/exp \
-  --use-averaged-model false \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --epoch 99 \
-  --avg 1 \
-  --jit 1
-
-ls -lh $repo/exp/*.pt
-
-log "Decode with models exported by torch.jit.script()"
-
-./zipformer_mmi/jit_pretrained.py \
-  --bpe-model $repo/data/lang_bpe_500/bpe.model \
-  --nn-model-filename $repo/exp/cpu_jit.pt \
-  --lang-dir $repo/data/lang_bpe_500 \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-for method in 1best nbest nbest-rescoring-LG nbest-rescoring-3-gram nbest-rescoring-4-gram; do
-  log "$method"
-
-  ./zipformer_mmi/pretrained.py \
-    --method $method \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_bpe_500 \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p zipformer_mmi/exp
-  ln -s $PWD/$repo/exp/pretrained.pt zipformer_mmi/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh zipformer_mmi/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in 1best nbest nbest-rescoring-LG nbest-rescoring-3-gram nbest-rescoring-4-gram; do
-    log "Decoding with $method"
-
-    ./zipformer_mmi/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --use-averaged-model 0 \
-      --nbest-scale 1.2 \
-      --hp-scale 1.0 \
-      --max-duration $max_duration \
-      --lang-dir $repo/data/lang_bpe_500 \
-      --exp-dir zipformer_mmi/exp
-  done
-
-  rm zipformer_mmi/exp/*.pt
-fi

.github/scripts/run-multi-corpora-zipformer.sh

@@ -95,3 +95,41 @@ for method in modified_beam_search fast_beam_search; do
   $repo/test_wavs/DEV_T0000000001.wav \
   $repo/test_wavs/DEV_T0000000002.wav
 done
+
+rm -rf $repo
+
+cd ../../../egs/multi_zh_en/ASR
+log "==== Test icefall-asr-zipformer-multi-zh-en-2023-11-22 ===="
+repo_url=https://huggingface.co/zrjin/icefall-asr-zipformer-multi-zh-en-2023-11-22/
+
+log "Downloading pre-trained model from $repo_url"
+git lfs install
+git clone $repo_url
+repo=$(basename $repo_url)
+
+log "Display test files"
+tree $repo/
+ls -lh $repo/test_wavs/*.wav
+
+./zipformer/pretrained.py \
+  --checkpoint $repo/exp/pretrained.pt \
+  --bpe-model $repo/data/lang_bbpe_2000/bbpe.model \
+  --method greedy_search \
+$repo/test_wavs/_1634_210_2577_1_1525157964032_3712259_29.wav \
+$repo/test_wavs/_1634_210_2577_1_1525157964032_3712259_55.wav \
+$repo/test_wavs/_1634_210_2577_1_1525157964032_3712259_75.wav
+
+for method in modified_beam_search fast_beam_search; do
+  log "$method"
+
+  ./zipformer/pretrained.py \
+    --method $method \
+    --beam-size 4 \
+    --checkpoint $repo/exp/pretrained.pt \
+    --bpe-model $repo/data/lang_bbpe_2000/bbpe.model \
+  $repo/test_wavs/_1634_210_2577_1_1525157964032_3712259_29.wav \
+  $repo/test_wavs/_1634_210_2577_1_1525157964032_3712259_55.wav \
+  $repo/test_wavs/_1634_210_2577_1_1525157964032_3712259_75.wav
+done
+
+rm -rf $repo

.github/scripts/run-pre-trained-ctc.sh

@@ -1,240 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-pushd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/sherpa-onnx-zipformer-ctc-en-2023-10-02
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-log "CTC greedy search"
-
-./zipformer/onnx_pretrained_ctc.py \
-  --nn-model $repo/model.onnx \
-  --tokens $repo/tokens.txt \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-log "CTC H decoding"
-
-./zipformer/onnx_pretrained_ctc_H.py \
-  --nn-model $repo/model.onnx \
-  --tokens $repo/tokens.txt \
-  --H $repo/H.fst \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-log "CTC HL decoding"
-
-./zipformer/onnx_pretrained_ctc_HL.py \
-  --nn-model $repo/model.onnx \
-  --words $repo/words.txt \
-  --HL $repo/HL.fst \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-log "CTC HLG decoding"
-
-./zipformer/onnx_pretrained_ctc_HLG.py \
-  --nn-model $repo/model.onnx \
-  --words $repo/words.txt \
-  --HLG $repo/HLG.fst \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-rm -rf $repo
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-conformer-ctc-jit-bpe-500-2021-11-09
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-pushd $repo
-
-git lfs pull --include "exp/pretrained.pt"
-git lfs pull --include "data/lang_bpe_500/HLG.pt"
-git lfs pull --include "data/lang_bpe_500/L.pt"
-git lfs pull --include "data/lang_bpe_500/L_disambig.pt"
-git lfs pull --include "data/lang_bpe_500/Linv.pt"
-git lfs pull --include "data/lang_bpe_500/bpe.model"
-git lfs pull --include "data/lang_bpe_500/lexicon.txt"
-git lfs pull --include "data/lang_bpe_500/lexicon_disambig.txt"
-git lfs pull --include "data/lang_bpe_500/tokens.txt"
-git lfs pull --include "data/lang_bpe_500/words.txt"
-git lfs pull --include "data/lm/G_3_gram.fst.txt"
-
-popd
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-log "CTC decoding"
-
-./conformer_ctc/pretrained.py \
-  --method ctc-decoding \
-  --num-classes 500 \
-  --checkpoint $repo/exp/pretrained.pt \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "HLG decoding"
-
-./conformer_ctc/pretrained.py \
-  --method 1best \
-  --num-classes 500 \
-  --checkpoint $repo/exp/pretrained.pt \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --words-file $repo/data/lang_bpe_500/words.txt \
-  --HLG $repo/data/lang_bpe_500/HLG.pt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "CTC decoding on CPU with kaldi decoders using OpenFst"
-
-log "Exporting model with torchscript"
-
-pushd $repo/exp
-ln -s pretrained.pt epoch-99.pt
-popd
-
-./conformer_ctc/export.py \
-  --epoch 99 \
-  --avg 1 \
-  --exp-dir $repo/exp \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  --jit 1
-
-ls -lh $repo/exp
-
-
-log "Generating H.fst, HL.fst"
-
-./local/prepare_lang_fst.py  --lang-dir $repo/data/lang_bpe_500 --ngram-G $repo/data/lm/G_3_gram.fst.txt
-
-ls -lh $repo/data/lang_bpe_500
-
-log "Decoding with H on CPU with OpenFst"
-
-./conformer_ctc/jit_pretrained_decode_with_H.py \
-  --nn-model $repo/exp/cpu_jit.pt \
-  --H $repo/data/lang_bpe_500/H.fst \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "Decoding with HL on CPU with OpenFst"
-
-./conformer_ctc/jit_pretrained_decode_with_HL.py \
-  --nn-model $repo/exp/cpu_jit.pt \
-  --HL $repo/data/lang_bpe_500/HL.fst \
-  --words $repo/data/lang_bpe_500/words.txt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-log "Decoding with HLG on CPU with OpenFst"
-
-./conformer_ctc/jit_pretrained_decode_with_HLG.py \
-  --nn-model $repo/exp/cpu_jit.pt \
-  --HLG $repo/data/lang_bpe_500/HLG.fst \
-  --words $repo/data/lang_bpe_500/words.txt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav
-
-rm -rf $repo
-
-popd
-
-log "Test aishell"
-
-pushd egs/aishell/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall_asr_aishell_conformer_ctc
-log "Downloading pre-trained model from $repo_url"
-GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
-repo=$(basename $repo_url)
-pushd $repo
-
-git lfs pull --include "exp/pretrained.pt"
-git lfs pull --include "data/lang_char/H.fst"
-git lfs pull --include "data/lang_char/HL.fst"
-git lfs pull --include "data/lang_char/HLG.fst"
-
-popd
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-log "CTC decoding"
-
-log "Exporting model with torchscript"
-
-pushd $repo/exp
-ln -s pretrained.pt epoch-99.pt
-popd
-
-./conformer_ctc/export.py \
-  --epoch 99 \
-  --avg 1 \
-  --exp-dir $repo/exp \
-  --tokens $repo/data/lang_char/tokens.txt \
-  --jit 1
-
-ls -lh $repo/exp
-
-ls -lh $repo/data/lang_char
-
-log "Decoding with H on CPU with OpenFst"
-
-./conformer_ctc/jit_pretrained_decode_with_H.py \
-  --nn-model $repo/exp/cpu_jit.pt \
-  --H $repo/data/lang_char/H.fst \
-  --tokens $repo/data/lang_char/tokens.txt \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-log "Decoding with HL on CPU with OpenFst"
-
-./conformer_ctc/jit_pretrained_decode_with_HL.py \
-  --nn-model $repo/exp/cpu_jit.pt \
-  --HL $repo/data/lang_char/HL.fst \
-  --words $repo/data/lang_char/words.txt \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-log "Decoding with HLG on CPU with OpenFst"
-
-./conformer_ctc/jit_pretrained_decode_with_HLG.py \
-  --nn-model $repo/exp/cpu_jit.pt \
-  --HLG $repo/data/lang_char/HLG.fst \
-  --words $repo/data/lang_char/words.txt \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav
-
-rm -rf $repo

.github/scripts/run-pre-trained-transducer-stateless-librispeech-100h.sh

@@ -1,77 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-100h-transducer-stateless-multi-datasets-bpe-500-2022-02-21
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./transducer_stateless_multi_datasets/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./transducer_stateless_multi_datasets/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p transducer_stateless_multi_datasets/exp
-  ln -s $PWD/$repo/exp/pretrained.pt transducer_stateless_multi_datasets/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh transducer_stateless_multi_datasets/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./transducer_stateless_multi_datasets/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir transducer_stateless_multi_datasets/exp
-  done
-
-  rm transducer_stateless_multi_datasets/exp/*.pt
-fi

.github/scripts/run-pre-trained-transducer-stateless-librispeech-960h.sh

@@ -1,77 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-transducer-stateless-multi-datasets-bpe-500-2022-03-01
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./transducer_stateless_multi_datasets/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in modified_beam_search beam_search fast_beam_search; do
-  log "$method"
-
-  ./transducer_stateless_multi_datasets/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p transducer_stateless_multi_datasets/exp
-  ln -s $PWD/$repo/exp/pretrained.pt transducer_stateless_multi_datasets/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh transducer_stateless_multi_datasets/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./transducer_stateless_multi_datasets/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir transducer_stateless_multi_datasets/exp
-  done
-
-  rm transducer_stateless_multi_datasets/exp/*.pt
-fi

.github/scripts/run-pre-trained-transducer-stateless-modified-2-aishell.sh

@@ -1,48 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/aishell/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-aishell-transducer-stateless-modified-2-2022-03-01
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./transducer_stateless_modified-2/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_char \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-
-for method in modified_beam_search beam_search; do
-  log "$method"
-
-  ./transducer_stateless_modified-2/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_char \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done

.github/scripts/run-pre-trained-transducer-stateless-modified-aishell.sh

@@ -1,48 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/aishell/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-aishell-transducer-stateless-modified-2022-03-01
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./transducer_stateless_modified/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_char \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done
-
-for method in modified_beam_search beam_search; do
-  log "$method"
-
-  ./transducer_stateless_modified/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --lang-dir $repo/data/lang_char \
-    $repo/test_wavs/BAC009S0764W0121.wav \
-    $repo/test_wavs/BAC009S0764W0122.wav \
-    $repo/test_wavs/BAC009S0764W0123.wav
-done

.github/scripts/run-pre-trained-transducer-stateless.sh

@@ -1,77 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-transducer-stateless-bpe-500-2022-02-07
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-for sym in 1 2 3; do
-  log "Greedy search with --max-sym-per-frame $sym"
-
-  ./transducer_stateless/pretrained.py \
-    --method greedy_search \
-    --max-sym-per-frame $sym \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-for method in fast_beam_search modified_beam_search beam_search; do
-  log "$method"
-
-  ./transducer_stateless/pretrained.py \
-    --method $method \
-    --beam-size 4 \
-    --checkpoint $repo/exp/pretrained.pt \
-    --tokens $repo/data/lang_bpe_500/tokens.txt \
-    $repo/test_wavs/1089-134686-0001.wav \
-    $repo/test_wavs/1221-135766-0001.wav \
-    $repo/test_wavs/1221-135766-0002.wav
-done
-
-echo "GITHUB_EVENT_NAME: ${GITHUB_EVENT_NAME}"
-echo "GITHUB_EVENT_LABEL_NAME: ${GITHUB_EVENT_LABEL_NAME}"
-if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == x"run-decode"  ]]; then
-  mkdir -p transducer_stateless/exp
-  ln -s $PWD/$repo/exp/pretrained.pt transducer_stateless/exp/epoch-999.pt
-  ln -s $PWD/$repo/data/lang_bpe_500 data/
-
-  ls -lh data
-  ls -lh transducer_stateless/exp
-
-  log "Decoding test-clean and test-other"
-
-  # use a small value for decoding with CPU
-  max_duration=100
-
-  for method in greedy_search fast_beam_search modified_beam_search; do
-    log "Decoding with $method"
-
-    ./transducer_stateless/decode.py \
-      --decoding-method $method \
-      --epoch 999 \
-      --avg 1 \
-      --max-duration $max_duration \
-      --exp-dir transducer_stateless/exp
-  done
-
-  rm transducer_stateless/exp/*.pt
-fi

.github/scripts/run-pre-trained-transducer.sh

@@ -1,33 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-log() {
-  # This function is from espnet
-  local fname=${BASH_SOURCE[1]##*/}
-  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
-}
-
-cd egs/librispeech/ASR
-
-repo_url=https://huggingface.co/csukuangfj/icefall-asr-librispeech-transducer-bpe-500-2021-12-23
-
-log "Downloading pre-trained model from $repo_url"
-git lfs install
-git clone $repo_url
-repo=$(basename $repo_url)
-
-log "Display test files"
-tree $repo/
-ls -lh $repo/test_wavs/*.wav
-
-log "Beam search decoding"
-
-./transducer/pretrained.py \
-  --method beam_search \
-  --beam-size 4 \
-  --checkpoint $repo/exp/pretrained.pt \
-  --tokens $repo/data/lang_bpe_500/tokens.txt \
-  $repo/test_wavs/1089-134686-0001.wav \
-  $repo/test_wavs/1221-135766-0001.wav \
-  $repo/test_wavs/1221-135766-0002.wav

.github/scripts/run-wenetspeech-pruned-transducer-stateless2.sh

@@ -30,7 +30,7 @@ log "Test exporting to ONNX format"
 
 ./pruned_transducer_stateless2/export-onnx.py \
   --exp-dir $repo/exp \
-  --lang-dir $repo/data/lang_char \
+  --tokens $repo/data/lang_char/tokens.txt \
   --epoch 99 \
   --avg 1
 
@@ -38,14 +38,14 @@ log "Export to torchscript model"
 
 ./pruned_transducer_stateless2/export.py \
   --exp-dir $repo/exp \
-  --lang-dir $repo/data/lang_char \
+  --tokens $repo/data/lang_char/tokens.txt \
   --epoch 99 \
   --avg 1 \
   --jit 1
 
 ./pruned_transducer_stateless2/export.py \
   --exp-dir $repo/exp \
-  --lang-dir $repo/data/lang_char \
+  --tokens $repo/data/lang_char/tokens.txt \
   --epoch 99 \
   --avg 1 \
   --jit-trace 1

.github/scripts/yesno/ASR/run.sh

@@ -0,0 +1,86 @@
+#!/usr/bin/env bash
+
+set -ex
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+cd egs/yesno/ASR
+
+log "data preparation"
+./prepare.sh
+
+log "training"
+python3 ./tdnn/train.py
+
+log "decoding"
+python3 ./tdnn/decode.py
+
+log "export to pretrained.pt"
+
+python3 ./tdnn/export.py --epoch 14 --avg 2
+
+python3 ./tdnn/pretrained.py \
+  --checkpoint ./tdnn/exp/pretrained.pt \
+  --HLG ./data/lang_phone/HLG.pt \
+  --words-file ./data/lang_phone/words.txt \
+  download/waves_yesno/0_0_0_1_0_0_0_1.wav \
+  download/waves_yesno/0_0_1_0_0_0_1_0.wav
+
+log "Test exporting to torchscript"
+python3 ./tdnn/export.py --epoch 14 --avg 2 --jit 1
+
+python3 ./tdnn/jit_pretrained.py \
+  --nn-model ./tdnn/exp/cpu_jit.pt \
+  --HLG ./data/lang_phone/HLG.pt \
+  --words-file ./data/lang_phone/words.txt \
+  download/waves_yesno/0_0_0_1_0_0_0_1.wav \
+  download/waves_yesno/0_0_1_0_0_0_1_0.wav
+
+log "Test exporting to onnx"
+python3 ./tdnn/export_onnx.py --epoch 14 --avg 2
+
+log "Test float32 model"
+python3 ./tdnn/onnx_pretrained.py \
+  --nn-model ./tdnn/exp/model-epoch-14-avg-2.onnx \
+  --HLG ./data/lang_phone/HLG.pt \
+  --words-file ./data/lang_phone/words.txt \
+  download/waves_yesno/0_0_0_1_0_0_0_1.wav \
+  download/waves_yesno/0_0_1_0_0_0_1_0.wav
+
+log "Test int8 model"
+python3 ./tdnn/onnx_pretrained.py \
+  --nn-model ./tdnn/exp/model-epoch-14-avg-2.int8.onnx \
+  --HLG ./data/lang_phone/HLG.pt \
+  --words-file ./data/lang_phone/words.txt \
+  download/waves_yesno/0_0_0_1_0_0_0_1.wav \
+  download/waves_yesno/0_0_1_0_0_0_1_0.wav
+
+log "Test decoding with H"
+python3 ./tdnn/export.py --epoch 14 --avg 2 --jit 1
+
+python3 ./tdnn/jit_pretrained_decode_with_H.py \
+    --nn-model ./tdnn/exp/cpu_jit.pt \
+    --H ./data/lang_phone/H.fst \
+    --tokens ./data/lang_phone/tokens.txt \
+    ./download/waves_yesno/0_0_0_1_0_0_0_1.wav \
+    ./download/waves_yesno/0_0_1_0_0_0_1_0.wav \
+    ./download/waves_yesno/0_0_1_0_0_1_1_1.wav
+
+log "Test decoding with HL"
+python3 ./tdnn/export.py --epoch 14 --avg 2 --jit 1
+
+python3 ./tdnn/jit_pretrained_decode_with_HL.py \
+    --nn-model ./tdnn/exp/cpu_jit.pt \
+    --HL ./data/lang_phone/HL.fst \
+    --words ./data/lang_phone/words.txt \
+    ./download/waves_yesno/0_0_0_1_0_0_0_1.wav \
+    ./download/waves_yesno/0_0_1_0_0_0_1_0.wav \
+    ./download/waves_yesno/0_0_1_0_0_1_1_1.wav
+
+log "Show generated files"
+ls -lh tdnn/exp
+ls -lh data/lang_phone

.github/workflows/aishell.yml

@@ -0,0 +1,72 @@
+name: aishell
+
+on:
+  push:
+    branches:
+      - master
+
+  pull_request:
+    branches:
+      - master
+
+  workflow_dispatch:
+
+concurrency:
+  group: aishell-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  generate_build_matrix:
+    if: (github.repository_owner == 'csukuangfj' || github.repository_owner == 'k2-fsa') && (github.event.label.name == 'ready' || github.event_name == 'push' || github.event_name == 'aishell')
+
+    # see https://github.com/pytorch/pytorch/pull/50633
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - name: Generating build matrix
+        id: set-matrix
+        run: |
+          # outputting for debugging purposes
+          python ./.github/scripts/docker/generate_build_matrix.py
+          MATRIX=$(python ./.github/scripts/docker/generate_build_matrix.py)
+          echo "::set-output name=matrix::${MATRIX}"
+  aishell:
+    needs: generate_build_matrix
+    name: py${{ matrix.python-version }} torch${{ matrix.torch-version }} v${{ matrix.version }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        ${{ fromJson(needs.generate_build_matrix.outputs.matrix) }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Free space
+        shell: bash
+        run: |
+          df -h
+          rm -rf /opt/hostedtoolcache
+          df -h
+          echo "pwd: $PWD"
+          echo "github.workspace ${{ github.workspace }}"
+
+      - name: Run aishell tests
+        uses: addnab/docker-run-action@v3
+        with:
+            image: ghcr.io/${{ github.repository_owner }}/icefall:cpu-py${{ matrix.python-version }}-torch${{ matrix.torch-version }}-v${{ matrix.version }}
+            options: |
+              --volume ${{ github.workspace }}/:/icefall
+            shell: bash
+            run: |
+              export PYTHONPATH=/icefall:$PYTHONPATH
+              cd /icefall
+              git config --global --add safe.directory /icefall
+
+              .github/scripts/aishell/ASR/run.sh

.github/workflows/build-cpu-docker.yml

@@ -0,0 +1,81 @@
+name: build-cpu-docker
+on:
+  workflow_dispatch:
+
+concurrency:
+  group: build-cpu-docker-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  generate_build_matrix:
+    if: github.repository_owner == 'csukuangfj' || github.repository_owner == 'k2-fsa'
+    # see https://github.com/pytorch/pytorch/pull/50633
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - name: Generating build matrix
+        id: set-matrix
+        run: |
+          # outputting for debugging purposes
+          python ./.github/scripts/docker/generate_build_matrix.py
+          MATRIX=$(python ./.github/scripts/docker/generate_build_matrix.py)
+          echo "::set-output name=matrix::${MATRIX}"
+  build-cpu-docker:
+    needs: generate_build_matrix
+    name: py${{ matrix.python-version }} torch${{ matrix.torch-version }} v${{ matrix.version }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        ${{ fromJson(needs.generate_build_matrix.outputs.matrix) }}
+
+    steps:
+      # refer to https://github.com/actions/checkout
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Free space
+        shell: bash
+        run: |
+          df -h
+          rm -rf /opt/hostedtoolcache
+          df -h
+
+      - name: 'Login to GitHub Container Registry'
+        uses: docker/login-action@v2
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Build docker Image
+        shell: bash
+        run: |
+          cd .github/scripts/docker
+          torch_version=${{ matrix.torch-version }}
+          torchaudio_version=${{ matrix.torchaudio-version }}
+
+          echo "torch_version: $torch_version"
+          echo "torchaudio_version: $torchaudio_version"
+
+          version=${{ matrix.version }}
+
+          tag=ghcr.io/${{ github.repository_owner }}/icefall:cpu-py${{ matrix.python-version }}-torch${{ matrix.torch-version }}-v$version
+          echo "tag: $tag"
+
+          docker build \
+            -t $tag \
+            --build-arg PYTHON_VERSION=${{ matrix.python-version }} \
+            --build-arg TORCH_VERSION=$torch_version \
+            --build-arg TORCHAUDIO_VERSION=$torchaudio_version \
+            --build-arg K2_VERSION=${{ matrix.k2-version }} \
+            --build-arg KALDIFEAT_VERSION=${{ matrix.kaldifeat-version }} \
+            .
+
+          docker image ls
+          docker push $tag

.github/workflows/librispeech.yml

@@ -0,0 +1,71 @@
+name: librispeech
+on:
+  push:
+    branches:
+      - master
+
+  pull_request:
+    branches:
+      - master
+
+  workflow_dispatch:
+
+concurrency:
+  group: librispeech-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  generate_build_matrix:
+    if: github.repository_owner == 'csukuangfj' || github.repository_owner == 'k2-fsa'
+    # see https://github.com/pytorch/pytorch/pull/50633
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - name: Generating build matrix
+        id: set-matrix
+        run: |
+          # outputting for debugging purposes
+          python ./.github/scripts/docker/generate_build_matrix.py
+          MATRIX=$(python ./.github/scripts/docker/generate_build_matrix.py)
+          echo "::set-output name=matrix::${MATRIX}"
+  librispeech:
+    needs: generate_build_matrix
+    name: py${{ matrix.python-version }} torch${{ matrix.torch-version }} v${{ matrix.version }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        ${{ fromJson(needs.generate_build_matrix.outputs.matrix) }}
+
+    steps:
+      # refer to https://github.com/actions/checkout
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Free space
+        shell: bash
+        run: |
+          df -h
+          rm -rf /opt/hostedtoolcache
+          df -h
+          echo "pwd: $PWD"
+          echo "github.workspace ${{ github.workspace }}"
+
+      - name: Test zipformer/train.py with LibriSpeech
+        uses: addnab/docker-run-action@v3
+        with:
+            image: ghcr.io/${{ github.repository_owner }}/icefall:cpu-py${{ matrix.python-version }}-torch${{ matrix.torch-version }}-v${{ matrix.version }}
+            options: |
+              --volume ${{ github.workspace }}/:/icefall
+            shell: bash
+            run: |
+              export PYTHONPATH=/icefall:$PYTHONPATH
+              cd /icefall
+              git config --global --add safe.directory /icefall
+
+              .github/scripts/librispeech/ASR/run.sh

.github/workflows/multi-zh-hans.yml

@@ -1,45 +1,30 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
+name: run-multi-zh-hans
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-transducer
 
 on:
   push:
     branches:
       - master
-  pull_request:
+
-    types: [labeled]
+  workflow_dispatch:
 
 concurrency:
-  group: run_pre_trained_transducer-${{ github.ref }}
+  group: run-multi-zh-hans-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: write
+
 jobs:
-  run_pre_trained_transducer:
+  multi-zh-hans:
-    if: github.event.label.name == 'ready' || github.event_name == 'push'
     runs-on: ${{ matrix.os }}
     strategy:
+      fail-fast: false
       matrix:
         os: [ubuntu-latest]
         python-version: [3.8]
 
-      fail-fast: false
-
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
 
@@ -68,13 +53,27 @@ jobs:
         if: steps.my-cache.outputs.cache-hit != 'true'
         shell: bash
         run: |
-          make -j2 _kaldifeat
+          .github/scripts/install-kaldifeat.sh
 
-      - name: Inference with pre-trained model
+      - name: export-model
         shell: bash
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
           sudo apt-get -qq install git-lfs tree
           export PYTHONPATH=$PWD:$PYTHONPATH
           export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
           export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-          .github/scripts/run-pre-trained-transducer.sh
+
+          .github/scripts/multi-zh-hans.sh
+          ls -lh
+
+      - name: upload model to https://github.com/k2-fsa/sherpa-onnx
+        uses: svenstaro/upload-release-action@v2
+        with:
+          file_glob: true
+          file: ./*.tar.bz2
+          overwrite: true
+          repo_name: k2-fsa/sherpa-onnx
+          repo_token: ${{ secrets.UPLOAD_GH_SHERPA_ONNX_TOKEN }}
+          tag: asr-models

.github/workflows/run-aishell-2022-06-20.yml

@@ -1,123 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-aishell-2022-06-20
-# pruned RNN-T + reworked model with random combiner
-# https://huggingface.co/csukuangfj/icefall-aishell-pruned-transducer-stateless3-2022-06-20
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_aishell_2022_06_20-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_aishell_2022_06_20:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh
-
-      - name: Display decoding results for aishell pruned_transducer_stateless3
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/aishell/ASR/
-          tree ./pruned_transducer_stateless3/exp
-
-          cd pruned_transducer_stateless3
-          echo "results for pruned_transducer_stateless3"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for dev" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for dev" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for dev" {} + | sort -n -k2
-
-      - name: Upload decoding results for aishell pruned_transducer_stateless3
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: aishell-torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless3-2022-06-20
-          path: egs/aishell/ASR/pruned_transducer_stateless3/exp/

.github/workflows/run-aishell-zipformer-2023-10-24.yml

@@ -1,95 +0,0 @@
-# Copyright      2023  Zengrui Jin (Xiaomi Corp.)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-aishell-zipformer-2023-10-24
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_aishell_zipformer_2023_10_24-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_aishell_zipformer_2023_10_24:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'zipformer' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-aishell-zipformer-2023-10-24.sh
-
-      

.github/workflows/run-gigaspeech-zipformer-2023-10-17.yml

@@ -21,6 +21,7 @@ on:
   push:
     branches:
       - master
+
   pull_request:
     types: [labeled]
 
@@ -33,6 +34,8 @@ on:
     # nightly build at 15:50 UTC time every day
     - cron: "50 15 * * *"
 
+  workflow_dispatch:
+
 concurrency:
   group: run_gigaspeech_2023_10_17_zipformer-${{ github.ref }}
   cancel-in-progress: true
@@ -85,6 +88,7 @@ jobs:
         env:
           GITHUB_EVENT_NAME: ${{ github.event_name }}
           GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
           mkdir -p egs/gigaspeech/ASR/data
           ln -sfv ~/tmp/fbank-libri egs/gigaspeech/ASR/data/fbank
@@ -97,6 +101,16 @@ jobs:
 
           .github/scripts/run-gigaspeech-zipformer-2023-10-17.sh
 
+      - name: upload model to https://github.com/k2-fsa/sherpa-onnx
+        uses: svenstaro/upload-release-action@v2
+        with:
+          file_glob: true
+          file: ./*.tar.bz2
+          overwrite: true
+          repo_name: k2-fsa/sherpa-onnx
+          repo_token: ${{ secrets.UPLOAD_GH_SHERPA_ONNX_TOKEN }}
+          tag: asr-models
+
       - name: Display decoding results for gigaspeech zipformer
         if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
         shell: bash

.github/workflows/run-librispeech-2022-03-12.yml

@@ -1,159 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-03-12
-# stateless transducer + k2 pruned rnnt-loss
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_03_12-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_03_12:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless-2022-03-12.sh
-
-      - name: Display decoding results for pruned_transducer_stateless
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless/exp
-
-          cd pruned_transducer_stateless
-          echo "results for pruned_transducer_stateless"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for pruned_transducer_stateless
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless-2022-03-12
-          path: egs/librispeech/ASR/pruned_transducer_stateless/exp/

.github/workflows/run-librispeech-2022-04-29.yml

@@ -1,185 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-04-29
-# stateless pruned transducer (reworked model) + giga speech
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_04_29-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_04_29:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless2-2022-04-29.sh
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless3-2022-04-29.sh
-
-      - name: Display decoding results for pruned_transducer_stateless2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR
-          tree pruned_transducer_stateless2/exp
-          cd pruned_transducer_stateless2/exp
-          echo "===greedy search==="
-          find greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Display decoding results for pruned_transducer_stateless3
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR
-          tree pruned_transducer_stateless3/exp
-          cd pruned_transducer_stateless3/exp
-          echo "===greedy search==="
-          find greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for pruned_transducer_stateless2
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless2-2022-04-29
-          path: egs/librispeech/ASR/pruned_transducer_stateless2/exp/
-
-      - name: Upload decoding results for pruned_transducer_stateless3
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless3-2022-04-29
-          path: egs/librispeech/ASR/pruned_transducer_stateless3/exp/

.github/workflows/run-librispeech-2022-05-13.yml

@@ -1,159 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-05-13
-# stateless transducer + k2 pruned rnnt-loss + deeper model
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_05_13-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_05_13:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless5-2022-05-13.sh
-
-      - name: Display decoding results for librispeech pruned_transducer_stateless5
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless5/exp
-
-          cd pruned_transducer_stateless5
-          echo "results for pruned_transducer_stateless5"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech pruned_transducer_stateless5
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless5-2022-05-13
-          path: egs/librispeech/ASR/pruned_transducer_stateless5/exp/

.github/workflows/run-librispeech-2022-11-11-stateless7.yml

@@ -1,159 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-11-11-stateless7
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_11_11_zipformer-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_11_11_zipformer:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless7-2022-11-11.sh
-
-      - name: Display decoding results for librispeech pruned_transducer_stateless7
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless7/exp
-
-          cd pruned_transducer_stateless7
-          echo "results for pruned_transducer_stateless7"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech pruned_transducer_stateless7
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless7-2022-11-11
-          path: egs/librispeech/ASR/pruned_transducer_stateless7/exp/

.github/workflows/run-librispeech-2022-11-14-stateless8.yml

@@ -1,159 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-11-14-stateless8
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_11_14_zipformer_stateless8-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_11_14_zipformer_stateless8:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless8-2022-11-14.sh
-
-      - name: Display decoding results for librispeech pruned_transducer_stateless8
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless8/exp
-
-          cd pruned_transducer_stateless8
-          echo "results for pruned_transducer_stateless8"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech pruned_transducer_stateless8
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless8-2022-11-14
-          path: egs/librispeech/ASR/pruned_transducer_stateless8/exp/

.github/workflows/run-librispeech-2022-12-01-stateless7-ctc.yml

@@ -1,163 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-12-01-stateless7-ctc
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-jobs:
-  run_librispeech_2022_11_11_zipformer:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-2022-12-01.sh
-
-      - name: Display decoding results for librispeech pruned_transducer_stateless7_ctc
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless7_ctc/exp
-
-          cd pruned_transducer_stateless7_ctc
-          echo "results for pruned_transducer_stateless7_ctc"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===ctc decoding==="
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===1best==="
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech pruned_transducer_stateless7_ctc
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless7-ctc-2022-12-01
-          path: egs/librispeech/ASR/pruned_transducer_stateless7_ctc/exp/

.github/workflows/run-librispeech-2022-12-08-zipformer-mmi.yml

@@ -1,167 +0,0 @@
-# Copyright      2022  Zengwei Yao
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-12-08-zipformer-mmi
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_12_08_zipformer-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_12_08_zipformer:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-zipformer-mmi-2022-12-08.sh
-
-      - name: Display decoding results for librispeech zipformer-mmi
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./zipformer-mmi/exp
-
-          cd zipformer-mmi
-          echo "results for zipformer-mmi"
-          echo "===1best==="
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===nbest==="
-          find exp/nbest -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/nbest -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===nbest-rescoring-LG==="
-          find exp/nbest-rescoring-LG -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/nbest-rescoring-LG -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===nbest-rescoring-3-gram==="
-          find exp/nbest-rescoring-3-gram -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/nbest-rescoring-3-gram -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===nbest-rescoring-4-gram==="
-          find exp/nbest-rescoring-4-gram -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/nbest-rescoring-4-gram -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech zipformer-mmi
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-zipformer_mmi-2022-12-08
-          path: egs/librispeech/ASR/zipformer_mmi/exp/

.github/workflows/run-librispeech-2022-12-29-stateless7-streaming.yml

@@ -1,172 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-12-29-stateless7-streaming
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_12_29_zipformer_streaming-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_12_29_zipformer_streaming:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event.label.name == 'streaming-zipformer' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless7-streaming-2022-12-29.sh
-
-      - name: Display decoding results for librispeech pruned_transducer_stateless7_streaming
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless7_streaming/exp
-
-          cd pruned_transducer_stateless7_streaming
-          echo "results for pruned_transducer_stateless7_streaming"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===streaming greedy search==="
-          find exp/streaming/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/streaming/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===streaming fast_beam_search==="
-          find exp/streaming/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/streaming/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===streaming modified beam search==="
-          find exp/streaming/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/streaming/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-
-      - name: Upload decoding results for librispeech pruned_transducer_stateless7_streaming
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless7-streaming-2022-12-29
-          path: egs/librispeech/ASR/pruned_transducer_stateless7_streaming/exp/

.github/workflows/run-librispeech-2023-01-29-stateless7-ctc-bs.yml

@@ -1,163 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2023-01-29-stateless7-ctc-bs
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-jobs:
-  run_librispeech_2023_01_29_zipformer_ctc_bs:
-    if: github.event.label.name == 'run-decode' || github.event.label.name == 'blank-skip' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-bs-2023-01-29.sh
-
-      - name: Display decoding results for librispeech pruned_transducer_stateless7_ctc_bs
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless7_ctc_bs/exp
-
-          cd pruned_transducer_stateless7_ctc_bs
-          echo "results for pruned_transducer_stateless7_ctc_bs"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===ctc decoding==="
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===1best==="
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech pruned_transducer_stateless7_ctc_bs
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless7-ctc-bs-2023-01-29
-          path: egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/exp/

.github/workflows/run-librispeech-conformer-ctc3-2022-11-28.yml

@@ -1,155 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-conformer-ctc3-2022-11-28
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_11_28_conformer_ctc3-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_11_28_conformer_ctc3:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-conformer-ctc3-2022-11-28.sh
-
-      - name: Display decoding results for librispeech conformer_ctc3
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./conformer_ctc3/exp
-
-          cd conformer_ctc3
-          echo "results for conformer_ctc3"
-          echo "===ctc-decoding==="
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===1best==="
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech conformer_ctc3
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-conformer_ctc3-2022-11-28
-          path: egs/librispeech/ASR/conformer_ctc3/exp/

.github/workflows/run-librispeech-pruned-transducer-stateless3-2022-05-13.yml

@@ -1,157 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-pruned-transducer-stateless3-2022-05-13
-# stateless pruned transducer (reworked model) + giga speech
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_pruned_transducer_stateless3_2022_05_13-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_pruned_transducer_stateless3_2022_05_13:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-pruned-transducer-stateless3-2022-05-13.sh
-
-      - name: Display decoding results for pruned_transducer_stateless3
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR
-          tree pruned_transducer_stateless3/exp
-          cd pruned_transducer_stateless3/exp
-          echo "===greedy search==="
-          find greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for pruned_transducer_stateless3
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless3-2022-04-29
-          path: egs/librispeech/ASR/pruned_transducer_stateless3/exp/

.github/workflows/run-librispeech-streaming-transducer-stateless2-2022-06-26.yml

@@ -1,159 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-streaming-2022-06-26
-# streaming conformer stateless transducer2
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_streaming_2022_06_26-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_streaming_2022_06_26:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-streaming-pruned-transducer-stateless2-2022-06-26.sh
-
-      - name: Display decoding results
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./pruned_transducer_stateless2/exp
-
-          cd pruned_transducer_stateless2
-          echo "results for pruned_transducer_stateless2"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified_beam_search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for pruned_transducer_stateless2
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-pruned_transducer_stateless2-2022-06-26
-          path: egs/librispeech/ASR/pruned_transducer_stateless2/exp/

.github/workflows/run-librispeech-streaming-zipformer-2023-05-18.yml

@@ -1,174 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-streaming-zipformer-2023-05-18
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2023_05_18_streaming_zipformer-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2023_05_18_streaming_zipformer:
-    if: github.event.label.name == 'zipformer' ||github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-streaming-zipformer-2023-05-18.sh
-
-      - name: Display decoding results for librispeech zipformer
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./zipformer/exp
-
-          cd zipformer
-
-          echo "results for zipformer, simulated streaming decoding"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "results for zipformer, chunk-wise streaming decoding"
-          echo "===greedy search==="
-          find exp/streaming/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/streaming/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/streaming/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/streaming/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/streaming/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/streaming/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-
-      - name: Upload decoding results for librispeech zipformer
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-zipformer-2022-11-11
-          path: egs/librispeech/ASR/zipformer/exp/

.github/workflows/run-librispeech-transducer-stateless2-2022-04-19.yml

@@ -1,159 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-2022-04-19
-# stateless transducer + torchaudio rnn-t loss
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2022_04_19-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2022_04_19:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-transducer-stateless2-2022-04-19.sh
-
-      - name: Display decoding results
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./transducer_stateless2/exp
-
-          cd transducer_stateless2
-          echo "results for transducer_stateless2"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified_beam_search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for transducer_stateless2
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-transducer_stateless2-2022-04-19
-          path: egs/librispeech/ASR/transducer_stateless2/exp/

.github/workflows/run-librispeech-zipformer-2023-05-18.yml

@@ -1,159 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-zipformer-2023-05-18
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2023_05_18_zipformer-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2023_05_18_zipformer:
-    if: github.event.label.name == 'zipformer' ||github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-zipformer-2023-05-18.sh
-
-      - name: Display decoding results for librispeech zipformer
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./zipformer/exp
-
-          cd zipformer
-          echo "results for zipformer"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech zipformer
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-zipformer-2022-11-11
-          path: egs/librispeech/ASR/zipformer/exp/

.github/workflows/run-librispeech-zipformer-ctc-2023-06-14.yml

@@ -1,155 +0,0 @@
-# Copyright      2022  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-librispeech-zipformer-ctc-2023-06-14
-# zipformer
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_librispeech_2023_06_14_zipformer-ctc-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_librispeech_2023_06_14_zipformer_ctc:
-    if: github.event.label.name == 'zipformer' ||github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-librispeech-zipformer-ctc-2023-06-14.sh
-
-      - name: Display decoding results for librispeech zipformer
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./zipformer/exp
-
-          cd zipformer
-          echo "results for zipformer"
-          echo "===ctc-decoding==="
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/ctc-decoding -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===1best==="
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/1best -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for librispeech zipformer
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-zipformer-2022-11-11
-          path: egs/librispeech/ASR/zipformer/exp/

.github/workflows/run-multi-corpora-zipformer.yml

@@ -14,7 +14,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-name: run-multi-zh_hans-zipformer
+name: run-multi-corpora-zipformer
 
 on:
   push:
@@ -24,12 +24,12 @@ on:
     types: [labeled]
 
 concurrency:
-  group: run_multi-zh_hans_zipformer-${{ github.ref }}
+  group: run_multi-corpora_zipformer-${{ github.ref }}
   cancel-in-progress: true
 
 jobs:
-  run_multi-zh_hans_zipformer:
+  run_multi-corpora_zipformer:
-    if: github.event.label.name == 'onnx' || github.event.label.name == 'ready' || github.event_name == 'push' || github.event.label.name == 'multi-zh_hans' || github.event.label.name == 'zipformer'
+    if: github.event.label.name == 'onnx' || github.event.label.name == 'ready' || github.event_name == 'push' || github.event.label.name == 'multi-zh_hans' || github.event.label.name == 'zipformer' || github.event.label.name == 'multi-corpora'
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
@@ -81,4 +81,4 @@ jobs:
           export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
           export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
 
-          .github/scripts/run-multi-zh_hans-zipformer.sh
+          .github/scripts/run-multi-corpora-zipformer.sh

.github/workflows/run-pretrained-ctc.yml

@@ -1,87 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-ctc
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  workflow_dispatch:
-    inputs:
-      test-run:
-        description: 'Test (y/n)?'
-        required: true
-        default: 'y'
-
-concurrency:
-  group: run_pre_trained_ctc-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_pre_trained_ctc:
-    if: github.event.label.name == 'ready' || github.event_name == 'push' || github.event.inputs.test-run == 'y' || github.event.label.name == 'ctc'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        run: |
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-          .github/scripts/run-pre-trained-ctc.sh

.github/workflows/run-pretrained-transducer-stateless-librispeech-100h.yml

@@ -1,158 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-trandsucer-stateless-multi-datasets-librispeech-100h
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_pre_trained_transducer_stateless_multi_datasets_librispeech_100h-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_pre_trained_transducer_stateless_multi_datasets_librispeech_100h:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-pre-trained-transducer-stateless-librispeech-100h.sh
-
-      - name: Display decoding results for transducer_stateless_multi_datasets
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./transducer_stateless_multi_datasets/exp
-
-          cd transducer_stateless_multi_datasets
-          echo "results for transducer_stateless_multi_datasets"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for transducer_stateless_multi_datasets
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-transducer_stateless_multi_datasets-100h-2022-02-21
-          path: egs/librispeech/ASR/transducer_stateless_multi_datasets/exp/

.github/workflows/run-pretrained-transducer-stateless-librispeech-multi-datasets.yml

@@ -1,158 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-trandsucer-stateless-multi-datasets-librispeech-960h
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_pre_trained_transducer_stateless_multi_datasets_librispeech_960h-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_pre_trained_transducer_stateless_multi_datasets_librispeech_960h:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-pre-trained-transducer-stateless-librispeech-960h.sh
-
-      - name: Display decoding results for transducer_stateless_multi_datasets
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./transducer_stateless_multi_datasets/exp
-
-          cd transducer_stateless_multi_datasets
-          echo "results for transducer_stateless_multi_datasets"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for transducer_stateless_multi_datasets
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-transducer_stateless_multi_datasets-100h-2022-03-01
-          path: egs/librispeech/ASR/transducer_stateless_multi_datasets/exp/

.github/workflows/run-pretrained-transducer-stateless-modified-2-aishell.yml

@@ -1,80 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-trandsucer-stateless-modified-2-aishell
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-concurrency:
-  group: run_pre_trained_transducer_stateless_modified_2_aishell-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_pre_trained_transducer_stateless_modified_2_aishell:
-    if: github.event.label.name == 'ready' || github.event_name == 'push'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        run: |
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-          .github/scripts/run-pre-trained-transducer-stateless-modified-2-aishell.sh

.github/workflows/run-pretrained-transducer-stateless-modified-aishell.yml

@@ -1,80 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-trandsucer-stateless-modified-aishell
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-concurrency:
-  group: run_pre_trained_transducer_stateless_modified_aishell-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_pre_trained_transducer_stateless_modified_aishell:
-    if: github.event.label.name == 'ready' || github.event_name == 'push'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        run: |
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-          .github/scripts/run-pre-trained-transducer-stateless-modified-aishell.sh

.github/workflows/run-pretrained-transducer-stateless.yml

@@ -1,158 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-pre-trained-transducer-stateless
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    types: [labeled]
-
-  schedule:
-    # minute (0-59)
-    # hour (0-23)
-    # day of the month (1-31)
-    # month (1-12)
-    # day of the week (0-6)
-    # nightly build at 15:50 UTC time every day
-    - cron: "50 15 * * *"
-
-concurrency:
-  group: run_pre_trained_transducer_stateless-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run_pre_trained_transducer_stateless:
-    if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event_name == 'push' || github.event_name == 'schedule'
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: [3.8]
-
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-      - name: Cache kaldifeat
-        id: my-cache
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/kaldifeat
-          key: cache-tmp-${{ matrix.python-version }}-2023-05-22
-
-      - name: Install kaldifeat
-        if: steps.my-cache.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/install-kaldifeat.sh
-
-      - name: Cache LibriSpeech test-clean and test-other datasets
-        id: libri-test-clean-and-test-other-data
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/download
-          key: cache-libri-test-clean-and-test-other
-
-      - name: Download LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-data.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/download-librispeech-test-clean-and-test-other-dataset.sh
-
-      - name: Prepare manifests for LibriSpeech test-clean and test-other
-        shell: bash
-        run: |
-          .github/scripts/prepare-librispeech-test-clean-and-test-other-manifests.sh
-
-      - name: Cache LibriSpeech test-clean and test-other fbank features
-        id: libri-test-clean-and-test-other-fbank
-        uses: actions/cache@v2
-        with:
-          path: |
-            ~/tmp/fbank-libri
-          key: cache-libri-fbank-test-clean-and-test-other-v2
-
-      - name: Compute fbank for LibriSpeech test-clean and test-other
-        if: steps.libri-test-clean-and-test-other-fbank.outputs.cache-hit != 'true'
-        shell: bash
-        run: |
-          .github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh
-
-      - name: Inference with pre-trained model
-        shell: bash
-        env:
-          GITHUB_EVENT_NAME: ${{ github.event_name }}
-          GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }}
-        run: |
-          mkdir -p egs/librispeech/ASR/data
-          ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank
-          ls -lh egs/librispeech/ASR/data/*
-
-          sudo apt-get -qq install git-lfs tree
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH
-          export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH
-
-          .github/scripts/run-pre-trained-transducer-stateless.sh
-
-      - name: Display decoding results for transducer_stateless
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        shell: bash
-        run: |
-          cd egs/librispeech/ASR/
-          tree ./transducer_stateless/exp
-
-          cd transducer_stateless
-          echo "results for transducer_stateless"
-          echo "===greedy search==="
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/greedy_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===fast_beam_search==="
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/fast_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-          echo "===modified beam search==="
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-clean" {} + | sort -n -k2
-          find exp/modified_beam_search -name "log-*" -exec grep -n --color "best for test-other" {} + | sort -n -k2
-
-      - name: Upload decoding results for transducer_stateless
-        uses: actions/upload-artifact@v2
-        if: github.event_name == 'schedule' || github.event.label.name == 'run-decode'
-        with:
-          name: torch-${{ matrix.torch }}-python-${{ matrix.python-version }}-ubuntu-latest-cpu-transducer_stateless-2022-02-07
-          path: egs/librispeech/ASR/transducer_stateless/exp/

.github/workflows/run-yesno-recipe.yml

@@ -1,185 +0,0 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-name: run-yesno-recipe
-
-on:
-  push:
-    branches:
-      - master
-  pull_request:
-    branches:
-      - master
-
-concurrency:
-  group: run-yesno-recipe-${{ github.ref }}
-  cancel-in-progress: true
-
-jobs:
-  run-yesno-recipe:
-    runs-on: ${{ matrix.os }}
-    strategy:
-      matrix:
-        # os: [ubuntu-latest, macos-10.15]
-        # TODO: enable macOS for CPU testing
-        os: [ubuntu-latest]
-        python-version: [3.8]
-      fail-fast: false
-
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          fetch-depth: 0
-
-      - name: Setup Python ${{ matrix.python-version }}
-        uses: actions/setup-python@v2
-        with:
-          python-version: ${{ matrix.python-version }}
-          cache: 'pip'
-          cache-dependency-path: '**/requirements-ci.txt'
-
-      - name: Install libnsdfile and libsox
-        if: startsWith(matrix.os, 'ubuntu')
-        run: |
-          sudo apt update
-          sudo apt install -q -y libsndfile1-dev libsndfile1 ffmpeg
-          sudo apt install -q -y --fix-missing sox libsox-dev libsox-fmt-all
-
-      - name: Install Python dependencies
-        run: |
-          grep -v '^#' ./requirements-ci.txt  | xargs -n 1 -L 1 pip install
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-          pip install --no-deps --force-reinstall k2==1.24.4.dev20231021+cpu.torch1.13.1 -f https://k2-fsa.github.io/k2/cpu.html
-          pip install kaldifeat==1.25.1.dev20231022+cpu.torch1.13.1 -f https://csukuangfj.github.io/kaldifeat/cpu.html
-
-      - name: Run yesno recipe
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          echo $PYTHONPATH
-
-          cd egs/yesno/ASR
-          ./prepare.sh
-          python3 ./tdnn/train.py
-          python3 ./tdnn/decode.py
-
-      - name: Test exporting to pretrained.pt
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          echo $PYTHONPATH
-
-          cd egs/yesno/ASR
-          python3 ./tdnn/export.py --epoch 14 --avg 2
-
-          python3 ./tdnn/pretrained.py \
-            --checkpoint ./tdnn/exp/pretrained.pt \
-            --HLG ./data/lang_phone/HLG.pt \
-            --words-file ./data/lang_phone/words.txt \
-            download/waves_yesno/0_0_0_1_0_0_0_1.wav \
-            download/waves_yesno/0_0_1_0_0_0_1_0.wav
-
-      - name: Test exporting to torchscript
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          echo $PYTHONPATH
-
-          cd egs/yesno/ASR
-          python3 ./tdnn/export.py --epoch 14 --avg 2 --jit 1
-
-          python3 ./tdnn/jit_pretrained.py \
-            --nn-model ./tdnn/exp/cpu_jit.pt \
-            --HLG ./data/lang_phone/HLG.pt \
-            --words-file ./data/lang_phone/words.txt \
-            download/waves_yesno/0_0_0_1_0_0_0_1.wav \
-            download/waves_yesno/0_0_1_0_0_0_1_0.wav
-
-      - name: Test exporting to onnx
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          echo $PYTHONPATH
-
-          cd egs/yesno/ASR
-          python3 ./tdnn/export_onnx.py --epoch 14 --avg 2
-
-          echo "Test float32 model"
-          python3 ./tdnn/onnx_pretrained.py \
-            --nn-model ./tdnn/exp/model-epoch-14-avg-2.onnx \
-            --HLG ./data/lang_phone/HLG.pt \
-            --words-file ./data/lang_phone/words.txt \
-            download/waves_yesno/0_0_0_1_0_0_0_1.wav \
-            download/waves_yesno/0_0_1_0_0_0_1_0.wav
-
-
-          echo "Test int8 model"
-          python3 ./tdnn/onnx_pretrained.py \
-            --nn-model ./tdnn/exp/model-epoch-14-avg-2.int8.onnx \
-            --HLG ./data/lang_phone/HLG.pt \
-            --words-file ./data/lang_phone/words.txt \
-            download/waves_yesno/0_0_0_1_0_0_0_1.wav \
-            download/waves_yesno/0_0_1_0_0_0_1_0.wav
-
-      - name: Test decoding with H
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          echo $PYTHONPATH
-
-          cd egs/yesno/ASR
-          python3 ./tdnn/export.py --epoch 14 --avg 2 --jit 1
-
-          python3 ./tdnn/jit_pretrained_decode_with_H.py \
-              --nn-model ./tdnn/exp/cpu_jit.pt \
-              --H ./data/lang_phone/H.fst \
-              --tokens ./data/lang_phone/tokens.txt \
-              ./download/waves_yesno/0_0_0_1_0_0_0_1.wav \
-              ./download/waves_yesno/0_0_1_0_0_0_1_0.wav \
-              ./download/waves_yesno/0_0_1_0_0_1_1_1.wav
-
-      - name: Test decoding with HL
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          export PYTHONPATH=$PWD:$PYTHONPATH
-          echo $PYTHONPATH
-
-          cd egs/yesno/ASR
-          python3 ./tdnn/export.py --epoch 14 --avg 2 --jit 1
-
-          python3 ./tdnn/jit_pretrained_decode_with_HL.py \
-              --nn-model ./tdnn/exp/cpu_jit.pt \
-              --HL ./data/lang_phone/HL.fst \
-              --words ./data/lang_phone/words.txt \
-              ./download/waves_yesno/0_0_0_1_0_0_0_1.wav \
-              ./download/waves_yesno/0_0_1_0_0_0_1_0.wav \
-              ./download/waves_yesno/0_0_1_0_0_1_1_1.wav
-
-      - name: Show generated files
-        shell: bash
-        working-directory: ${{github.workspace}}
-        run: |
-          cd egs/yesno/ASR
-          ls -lh tdnn/exp
-          ls -lh data/lang_phone

.github/workflows/test.yml

@@ -1,129 +1,109 @@
-# Copyright      2021  Fangjun Kuang (csukuangfj@gmail.com)
-
-# See ../../LICENSE for clarification regarding multiple authors
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
 name: test
 
 on:
   push:
     branches:
       - master
+
   pull_request:
     branches:
       - master
 
+  workflow_dispatch:
+
 concurrency:
   group: test-${{ github.ref }}
   cancel-in-progress: true
 
 jobs:
+  generate_build_matrix:
+    if: github.repository_owner == 'csukuangfj' || github.repository_owner == 'k2-fsa'
+    # see https://github.com/pytorch/pytorch/pull/50633
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - name: Generating build matrix
+        id: set-matrix
+        run: |
+          # outputting for debugging purposes
+          python ./.github/scripts/docker/generate_build_matrix.py
+          MATRIX=$(python ./.github/scripts/docker/generate_build_matrix.py)
+          echo "::set-output name=matrix::${MATRIX}"
   test:
-    runs-on: ${{ matrix.os }}
+    needs: generate_build_matrix
+    name: py${{ matrix.python-version }} torch${{ matrix.torch-version }} v${{ matrix.version }}
+    runs-on: ubuntu-latest
     strategy:
-      matrix:
-        os: [ubuntu-latest]
-        python-version: ["3.8"]
-        torch: ["1.13.0"]
-        torchaudio: ["0.13.0"]
-        k2-version: ["1.24.3.dev20230719"]
-
       fail-fast: false
+      matrix:
+        ${{ fromJson(needs.generate_build_matrix.outputs.matrix) }}
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
 
-      - name: Setup Python ${{ matrix.python-version }}
+      - name: Free space
-        uses: actions/setup-python@v1
-        with:
-          python-version: ${{ matrix.python-version }}
-
-      - name: Install libnsdfile and libsox
-        if: startsWith(matrix.os, 'ubuntu')
-        run: |
-          sudo apt update
-          sudo apt install -q -y libsndfile1-dev libsndfile1 ffmpeg
-          sudo apt install -q -y --fix-missing libsox-dev libsox-fmt-all
-
-      - name: Install Python dependencies
-        run: |
-          python3 -m pip install --upgrade pip pytest
-          # numpy 1.20.x does not support python 3.6
-          pip install numpy==1.19
-          pip install torch==${{ matrix.torch }}+cpu -f https://download.pytorch.org/whl/cpu/torch_stable.html
-          pip install torchaudio==${{ matrix.torchaudio }}+cpu -f https://download.pytorch.org/whl/cpu/torch_stable.html
-
-          pip install k2==${{ matrix.k2-version }}+cpu.torch${{ matrix.torch }} -f https://k2-fsa.github.io/k2/cpu.html
-          pip install git+https://github.com/lhotse-speech/lhotse
-          # icefall requirements
-          pip uninstall -y protobuf
-          pip install --no-binary protobuf protobuf==3.20.*
-
-          pip install kaldifst
-          pip install onnxruntime matplotlib
-          pip install -r requirements.txt
-
-      - name: Install graphviz
-        if: startsWith(matrix.os, 'ubuntu')
         shell: bash
         run: |
-          python3 -m pip install -qq graphviz
+          df -h
-          sudo apt-get -qq install graphviz
+          rm -rf /opt/hostedtoolcache
+          df -h
+          echo "pwd: $PWD"
+          echo "github.workspace ${{ github.workspace }}"
 
       - name: Run tests
-        if: startsWith(matrix.os, 'ubuntu')
+        uses: addnab/docker-run-action@v3
-        run: |
+        with:
-          ls -lh
+            image: ghcr.io/${{ github.repository_owner }}/icefall:cpu-py${{ matrix.python-version }}-torch${{ matrix.torch-version }}-v${{ matrix.version }}
-          export PYTHONPATH=$PWD:$PWD/lhotse:$PYTHONPATH
+            options: |
-          echo $PYTHONPATH
+              --volume ${{ github.workspace }}/:/icefall
-          pytest -v -s ./test
+            shell: bash
-          # runt tests for conformer ctc
+            run: |
-          cd egs/librispeech/ASR/conformer_ctc
+              export PYTHONPATH=/icefall:$PYTHONPATH
-          pytest -v -s
+              cd /icefall
+              git config --global --add safe.directory /icefall
 
-          cd ../pruned_transducer_stateless
+              pytest -v -s ./test
-          pytest -v -s
 
-          cd ../pruned_transducer_stateless2
+              # runt tests for conformer ctc
-          pytest -v -s
+              cd egs/librispeech/ASR/conformer_ctc
+              pytest -v -s
 
-          cd ../pruned_transducer_stateless3
+              cd ../pruned_transducer_stateless
-          pytest -v -s
+              pytest -v -s
 
-          cd ../pruned_transducer_stateless4
+              cd ../pruned_transducer_stateless2
-          pytest -v -s
+              pytest -v -s
 
-          echo $PYTHONPATH
+              cd ../pruned_transducer_stateless3
-          cd ../pruned_transducer_stateless7
+              pytest -v -s
-          pytest -v -s
 
-          cd ../transducer_stateless
+              cd ../pruned_transducer_stateless4
-          pytest -v -s
+              pytest -v -s
 
-          # cd ../transducer
+              echo $PYTHONPATH
-          # pytest -v -s
+              cd ../pruned_transducer_stateless7
+              pytest -v -s
 
-          cd ../transducer_stateless2
+              cd ../transducer_stateless
-          pytest -v -s
+              pytest -v -s
 
-          cd ../transducer_lstm
+              # cd ../transducer
-          pytest -v -s
+              # pytest -v -s
 
-          cd ../zipformer
+              cd ../transducer_stateless2
-          pytest -v -s
+              pytest -v -s
+
+              cd ../transducer_lstm
+              pytest -v -s
+
+              cd ../zipformer
+              pytest -v -s
 
       - uses: actions/upload-artifact@v2
         with:

.github/workflows/yesno.yml

@@ -0,0 +1,62 @@
+name: yesno
+
+on:
+  push:
+    branches:
+      - master
+
+  pull_request:
+    branches:
+      - master
+
+  workflow_dispatch:
+
+concurrency:
+  group: yesno-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  generate_build_matrix:
+    if: github.repository_owner == 'csukuangfj' || github.repository_owner == 'k2-fsa'
+    # see https://github.com/pytorch/pytorch/pull/50633
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+      - name: Generating build matrix
+        id: set-matrix
+        run: |
+          # outputting for debugging purposes
+          python ./.github/scripts/docker/generate_build_matrix.py
+          MATRIX=$(python ./.github/scripts/docker/generate_build_matrix.py)
+          echo "::set-output name=matrix::${MATRIX}"
+  yesno:
+    needs: generate_build_matrix
+    name: py${{ matrix.python-version }} torch${{ matrix.torch-version }} v${{ matrix.version }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        ${{ fromJson(needs.generate_build_matrix.outputs.matrix) }}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Run the yesno recipe
+        uses: addnab/docker-run-action@v3
+        with:
+            image: ghcr.io/${{ github.repository_owner }}/icefall:cpu-py${{ matrix.python-version }}-torch${{ matrix.torch-version }}-v${{ matrix.version }}
+            options: |
+              --volume ${{ github.workspace }}/:/icefall
+            shell: bash
+            run: |
+              export PYTHONPATH=/icefall:$PYTHONPATH
+              cd /icefall
+              git config --global --add safe.directory /icefall
+
+              .github/scripts/yesno/ASR/run.sh

README.md

@@ -2,46 +2,86 @@
 <img src="https://raw.githubusercontent.com/k2-fsa/icefall/master/docs/source/_static/logo.png" width=168>
 </div>
 
-## Introduction
+# Introduction
 
-icefall contains ASR recipes for various datasets
+The icefall project contains speech-related recipes for various datasets
-using <https://github.com/k2-fsa/k2>.
+using [k2-fsa](https://github.com/k2-fsa/k2) and [lhotse](https://github.com/lhotse-speech/lhotse).
 
-You can use <https://github.com/k2-fsa/sherpa> to deploy models
+You can use [sherpa](https://github.com/k2-fsa/sherpa), [sherpa-ncnn](https://github.com/k2-fsa/sherpa-ncnn) or [sherpa-onnx](https://github.com/k2-fsa/sherpa-onnx) for deployment with models
-trained with icefall.
+in icefall; these frameworks also support models not included in icefall; please refer to respective documents for more details.
 
 You can try pre-trained models from within your browser without the need
-to download or install anything by visiting <https://huggingface.co/spaces/k2-fsa/automatic-speech-recognition>
+to download or install anything by visiting this [huggingface space](https://huggingface.co/spaces/k2-fsa/automatic-speech-recognition).
-See <https://k2-fsa.github.io/icefall/huggingface/spaces.html> for more details.
+Please refer to [document](https://k2-fsa.github.io/icefall/huggingface/spaces.html) for more details.
 
-## Installation
+# Installation
 
-Please refer to <https://icefall.readthedocs.io/en/latest/installation/index.html>
+Please refer to [document](https://icefall.readthedocs.io/en/latest/installation/index.html)
 for installation.
 
-## Recipes
+# Recipes
 
-Please refer to <https://icefall.readthedocs.io/en/latest/recipes/index.html>
+Please refer to [document](https://icefall.readthedocs.io/en/latest/recipes/index.html)
-for more information.
+for more details.
 
-We provide the following recipes:
+## ASR: Automatic Speech Recognition
 
+### Supported Datasets
   - [yesno][yesno]
-  - [LibriSpeech][librispeech]
+  
-  - [GigaSpeech][gigaspeech]
+  - [Aidatatang_200zh][aidatatang_200zh]
-  - [AMI][ami]
   - [Aishell][aishell]
   - [Aishell2][aishell2]
   - [Aishell4][aishell4]
+  - [Alimeeting][alimeeting]
+  - [AMI][ami]
+  - [CommonVoice][commonvoice]
+  - [Corpus of Spontaneous Japanese][csj]
+  - [GigaSpeech][gigaspeech]
+  - [LibriCSS][libricss]
+  - [LibriSpeech][librispeech]
+  - [Libriheavy][libriheavy]
+  - [Multi-Dialect Broadcast News Arabic Speech Recognition][mgb2]
+  - [PeopleSpeech][peoplespeech]
+  - [SPGISpeech][spgispeech]
+  - [Switchboard][swbd]
   - [TIMIT][timit]
   - [TED-LIUM3][tedlium3]
-  - [Aidatatang_200zh][aidatatang_200zh]
-  - [WenetSpeech][wenetspeech]
-  - [Alimeeting][alimeeting]
-  - [Switchboard][swbd]
   - [TAL_CSASR][tal_csasr]
+  - [Voxpopuli][voxpopuli]
+  - [XBMU-AMDO31][xbmu-amdo31]
+  - [WenetSpeech][wenetspeech]
   
-### yesno
+More datasets will be added in the future.
+
+### Supported Models
+
+The [LibriSpeech][librispeech] recipe supports the most comprehensive set of models, you are welcome to try them out.
+
+#### CTC 
+  - TDNN LSTM CTC
+  - Conformer CTC
+  - Zipformer CTC
+
+#### MMI
+  - Conformer MMI
+  - Zipformer MMI
+
+#### Transducer
+  - Conformer-based Encoder
+  - LSTM-based Encoder
+  - Zipformer-based Encoder
+  - LSTM-based Predictor
+  - [Stateless Predictor](https://research.google/pubs/rnn-transducer-with-stateless-prediction-network/)
+
+#### Whisper
+  - [OpenAi Whisper](https://arxiv.org/abs/2212.04356) (We support fine-tuning on AiShell-1.)
+
+If you are willing to contribute to icefall, please refer to [contributing](https://icefall.readthedocs.io/en/latest/contributing/index.html) for more details.
+
+We would like to highlight the performance of some of the recipes here.
+
+### [yesno][yesno]
 
 This is the simplest ASR recipe in `icefall` and can be run on CPU.
 Training takes less than 30 seconds and gives you the following WER:
@@ -52,350 +92,264 @@ Training takes less than 30 seconds and gives you the following WER:
 We provide a Colab notebook for this recipe: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1tIjjzaJc3IvGyKiMCDWO-TSnBgkcuN3B?usp=sharing)
 
 
-### LibriSpeech
+### [LibriSpeech][librispeech]
 
-Please see <https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/RESULTS.md>
+Please see [RESULTS.md](https://github.com/k2-fsa/icefall/blob/master/egs/librispeech/ASR/RESULTS.md)
 for the **latest** results.
 
-We provide 5 models for this recipe:
+#### [Conformer CTC](https://github.com/k2-fsa/icefall/tree/master/egs/librispeech/ASR/conformer_ctc)
-
-- [conformer CTC model][LibriSpeech_conformer_ctc]
-- [TDNN LSTM CTC model][LibriSpeech_tdnn_lstm_ctc]
-- [Transducer: Conformer encoder + LSTM decoder][LibriSpeech_transducer]
-- [Transducer: Conformer encoder + Embedding decoder][LibriSpeech_transducer_stateless]
-- [Transducer: Zipformer encoder + Embedding decoder][LibriSpeech_zipformer]
-
-#### Conformer CTC Model
-
-The best WER we currently have is:
 
 |     | test-clean | test-other |
 |-----|------------|------------|
 | WER | 2.42       | 5.73       |
 
 
-We provide a Colab notebook to run a pre-trained conformer CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1huyupXAcHsUrKaWfI83iMEJ6J0Nh0213?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1huyupXAcHsUrKaWfI83iMEJ6J0Nh0213?usp=sharing)
 
-#### TDNN LSTM CTC Model
+#### [TDNN LSTM CTC](https://github.com/k2-fsa/icefall/tree/master/egs/librispeech/ASR/tdnn_lstm_ctc)
-
-The WER for this model is:
 
 |     | test-clean | test-other |
 |-----|------------|------------|
 | WER | 6.59       | 17.69      |
 
-We provide a Colab notebook to run a pre-trained TDNN LSTM CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1-iSfQMp2So-We_Uu49N4AAcMInB72u9z?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1-iSfQMp2So-We_Uu49N4AAcMInB72u9z?usp=sharing)
 
 
-#### Transducer: Conformer encoder + LSTM decoder
+#### [Transducer (Conformer Encoder + LSTM Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/librispeech/ASR/transducer)
 
-Using Conformer as encoder and LSTM as decoder.
+|               | test-clean | test-other |
+|---------------|------------|------------|
+| greedy_search | 3.07       | 7.51       |
 
-The best WER with greedy search is:
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1_u6yK9jDkPwG_NLrZMN2XK7Aeq4suMO2?usp=sharing)
 
-|     | test-clean | test-other |
+#### [Transducer (Conformer Encoder + Stateless Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/librispeech/ASR/transducer)
-|-----|------------|------------|
-| WER | 3.07       | 7.51       |
 
-We provide a Colab notebook to run a pre-trained RNN-T conformer model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1_u6yK9jDkPwG_NLrZMN2XK7Aeq4suMO2?usp=sharing)
+|                                       | test-clean | test-other |
-
+|---------------------------------------|------------|------------|
-#### Transducer: Conformer encoder + Embedding decoder
+| modified_beam_search (`beam_size=4`) | 2.56       | 6.27       |
-
-Using Conformer as encoder. The decoder consists of 1 embedding layer
-and 1 convolutional layer.
-
-The best WER using modified beam search with beam size 4 is:
-
-|     | test-clean | test-other |
-|-----|------------|------------|
-| WER | 2.56       | 6.27       |
-
-Note: No auxiliary losses are used in the training and no LMs are used
-in the decoding.
-
-We provide a Colab notebook to run a pre-trained transducer conformer + stateless decoder model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1CO1bXJ-2khDckZIW8zjOPHGSKLHpTDlp?usp=sharing)
 
 
-#### k2 pruned RNN-T
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1CO1bXJ-2khDckZIW8zjOPHGSKLHpTDlp?usp=sharing)
+
+
+#### [Transducer (Zipformer Encoder + Stateless Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/librispeech/ASR/zipformer)
+
+WER (modified_beam_search `beam_size=4` unless further stated) 
+
+1. LibriSpeech-960hr
 
 | Encoder         | Params | test-clean | test-other | epochs  | devices    |
 |-----------------|--------|------------|------------|---------|------------|
-| zipformer       | 65.5M  | 2.21       | 4.79       | 50      | 4 32G-V100 |
+| Zipformer       | 65.5M  | 2.21       | 4.79       | 50      | 4 32G-V100 |
-| zipformer-small | 23.2M  | 2.42       | 5.73       | 50      | 2 32G-V100 |
+| Zipformer-small | 23.2M  | 2.42       | 5.73       | 50      | 2 32G-V100 |
-| zipformer-large | 148.4M | 2.06       | 4.63       | 50      | 4 32G-V100 |
+| Zipformer-large | 148.4M | 2.06       | 4.63       | 50      | 4 32G-V100 |
-| zipformer-large | 148.4M | 2.00       | 4.38       | 174     | 8 80G-A100 |
+| Zipformer-large | 148.4M | 2.00       | 4.38       | 174     | 8 80G-A100 |
 
-Note: No auxiliary losses are used in the training and no LMs are used
+2. LibriSpeech-960hr + GigaSpeech
-in the decoding.
 
-#### k2 pruned RNN-T + GigaSpeech
+| Encoder         | Params | test-clean | test-other |
-
+|-----------------|--------|------------|------------|
-|     | test-clean | test-other |
+| Zipformer       | 65.5M   | 1.78       | 4.08       |
-|-----|------------|------------|
-| WER | 1.78       | 4.08       |
-
-Note: No auxiliary losses are used in the training and no LMs are used
-in the decoding.
-
-#### k2 pruned RNN-T + GigaSpeech + CommonVoice
-
-|     | test-clean | test-other |
-|-----|------------|------------|
-| WER | 1.90       | 3.98       |
-
-Note: No auxiliary losses are used in the training and no LMs are used
-in the decoding.
 
 
-### GigaSpeech
+3. LibriSpeech-960hr + GigaSpeech + CommonVoice
 
-We provide three models for this recipe:
+| Encoder         | Params | test-clean | test-other |
+|-----------------|--------|------------|------------|
+| Zipformer       | 65.5M   | 1.90       | 3.98       |
 
-- [Conformer CTC model][GigaSpeech_conformer_ctc]
-- [Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][GigaSpeech_pruned_transducer_stateless2].
-- [Transducer: Zipformer encoder + Embedding decoder][GigaSpeech_zipformer]
 
-#### Conformer CTC
+### [GigaSpeech][gigaspeech]
+
+#### [Conformer CTC](https://github.com/k2-fsa/icefall/tree/master/egs/gigaspeech/ASR/conformer_ctc)
 
 |     |  Dev  | Test  |
 |-----|-------|-------|
 | WER | 10.47 | 10.58 |
 
-#### Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss
+#### [Transducer (pruned_transducer_stateless2)](https://github.com/k2-fsa/icefall/tree/master/egs/gigaspeech/ASR/pruned_transducer_stateless2)
+
+Conformer Encoder + Stateless Predictor + k2 Pruned RNN-T Loss
 
 |                      |  Dev  | Test  |
 |----------------------|-------|-------|
-|    greedy search     | 10.51 | 10.73 |
+|    greedy_search     | 10.51 | 10.73 |
-|   fast beam search   | 10.50 | 10.69 |
+|   fast_beam_search   | 10.50 | 10.69 |
-| modified beam search | 10.40 | 10.51 |
+| modified_beam_search | 10.40 | 10.51 |
 
-#### Transducer: Zipformer encoder + Embedding decoder
+#### [Transducer (Zipformer Encoder + Stateless Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/gigaspeech/ASR/zipformer)
 
 |                      |  Dev  | Test  |
 |----------------------|-------|-------|
-|    greedy search     | 10.31 | 10.50 |
+|    greedy_search     | 10.31 | 10.50 |
-|   fast beam search   | 10.26 | 10.48 |
+|   fast_beam_search   | 10.26 | 10.48 |
-| modified beam search | 10.25 | 10.38 |
+| modified_beam_search | 10.25 | 10.38 |
 
 
-### Aishell
+### [Aishell][aishell]
 
-We provide three models for this recipe: [conformer CTC model][Aishell_conformer_ctc],
+#### [TDNN LSTM CTC](https://github.com/k2-fsa/icefall/tree/master/egs/aishell/ASR/tdnn_lstm_ctc)
-[TDNN LSTM CTC model][Aishell_tdnn_lstm_ctc], and [Transducer Stateless Model][Aishell_pruned_transducer_stateless7],
-
-#### Conformer CTC Model
-
-The best CER we currently have is:
-
-|     | test |
-|-----|------|
-| CER | 4.26 |
-
-#### TDNN LSTM CTC Model
-
-The CER for this model is:
 
 |     | test  |
 |-----|-------|
 | CER | 10.16 |
 
-We provide a Colab notebook to run a pre-trained TDNN LSTM CTC model:  [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1jbyzYq3ytm6j2nlEt-diQm-6QVWyDDEa?usp=sharing)
+We provide a Colab notebook to test the pre-trained model:  [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1jbyzYq3ytm6j2nlEt-diQm-6QVWyDDEa?usp=sharing)
 
-#### Transducer Stateless Model
+#### [Transducer (Conformer Encoder + Stateless Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/aishell/ASR/transducer_stateless)
-
-The best CER we currently have is:
 
 |     | test |
 |-----|------|
 | CER | 4.38 |
 
-We provide a Colab notebook to run a pre-trained TransducerStateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/14XaT2MhnBkK-3_RqqWq3K90Xlbin-GZC?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/14XaT2MhnBkK-3_RqqWq3K90Xlbin-GZC?usp=sharing)
+
+#### [Transducer (Zipformer Encoder + Stateless Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/aishell/ASR/zipformer)
+
+WER (modified_beam_search `beam_size=4`) 
+
+| Encoder         | Params | dev | test | epochs  |
+|-----------------|--------|-----|------|---------|
+| Zipformer       | 73.4M  | 4.13| 4.40 | 55      |
+| Zipformer-small | 30.2M  | 4.40| 4.67 | 55      |
+| Zipformer-large | 157.3M | 4.03| 4.28 | 56      |
 
 
-### Aishell2
+### [Aishell4][aishell4]
 
-We provide one model for this recipe: [Transducer Stateless Model][Aishell2_pruned_transducer_stateless5].
+#### [Transducer (pruned_transducer_stateless5)](https://github.com/k2-fsa/icefall/tree/master/egs/aishell4/ASR/pruned_transducer_stateless5)
-
-#### Transducer Stateless Model
-
-The best WER we currently have is:
-
-|     |   dev-ios  |  test-ios  |
-|-----|------------|------------|
-| WER |    5.32    |    5.56    |
-
-
-### Aishell4
-
-We provide one model for this recipe: [Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][Aishell4_pruned_transducer_stateless5].
-
-#### Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss (trained with all subsets)
-
-The best CER we currently have is:
 
+1 Trained with all subsets: 
 |     |   test     |
 |-----|------------|
 | CER |   29.08    |
 
-
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1z3lkURVv9M7uTiIgf3Np9IntMHEknaks?usp=sharing)
-We provide a Colab notebook to run a pre-trained Pruned Transducer Stateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1z3lkURVv9M7uTiIgf3Np9IntMHEknaks?usp=sharing)
 
 
-### TIMIT
+### [TIMIT][timit]
 
-We provide two models for this recipe: [TDNN LSTM CTC model][TIMIT_tdnn_lstm_ctc]
+#### [TDNN LSTM CTC](https://github.com/k2-fsa/icefall/tree/master/egs/timit/ASR/tdnn_lstm_ctc)
-and [TDNN LiGRU CTC model][TIMIT_tdnn_ligru_ctc].
 
-#### TDNN LSTM CTC Model
+|   |TEST|
-
+|---|----|
-The best PER we currently have is:
-
-||TEST|
-|--|--|
 |PER| 19.71% |
 
-We provide a Colab notebook to run a pre-trained TDNN LSTM CTC model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1Hs9DA4V96uapw_30uNp32OMJgkuR5VVd?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1Hs9DA4V96uapw_30uNp32OMJgkuR5VVd?usp=sharing)
 
-#### TDNN LiGRU CTC Model
+#### [TDNN LiGRU CTC](https://github.com/k2-fsa/icefall/tree/master/egs/timit/ASR/tdnn_ligru_ctc)
 
-The PER for this model is:
+|   |TEST|
-
+|---|----|
-||TEST|
-|--|--|
 |PER| 17.66% |
 
-We provide a Colab notebook to run a pre-trained TDNN LiGRU CTC model:  [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1z3lkURVv9M7uTiIgf3Np9IntMHEknaks?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1z3lkURVv9M7uTiIgf3Np9IntMHEknaks?usp=sharing)
 
 
-### TED-LIUM3
+### [TED-LIUM3][tedlium3]
 
-We provide two models for this recipe: [Transducer Stateless: Conformer encoder + Embedding decoder][TED-LIUM3_transducer_stateless] and [Pruned Transducer Stateless: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][TED-LIUM3_pruned_transducer_stateless].
+#### [Transducer (Conformer Encoder + Stateless Predictor)](https://github.com/k2-fsa/icefall/tree/master/egs/tedlium3/ASR/transducer_stateless)
 
-#### Transducer Stateless:  Conformer encoder + Embedding decoder
+|                                      |  dev  |  test  |
-
+|--------------------------------------|-------|--------|
-The best WER using modified beam search with beam size 4 is:
+| modified_beam_search (`beam_size=4`) |  6.91 |  6.33  |
-
-|     |  dev  |  test  |
-|-----|-------|--------|
-| WER |  6.91 |  6.33  |
-
-Note: No auxiliary losses are used in the training and no LMs are used in the decoding.
-
-We provide a Colab notebook to run a pre-trained Transducer Stateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1MmY5bBxwvKLNT4A2DJnwiqRXhdchUqPN?usp=sharing)
-
-#### Pruned Transducer Stateless: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss
-
-The best WER using modified beam search with beam size 4 is:
-
-|     |  dev  |  test  |
-|-----|-------|--------|
-| WER |  6.77 |  6.14  |
-
-We provide a Colab notebook to run a pre-trained Pruned Transducer Stateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1je_1zGrOkGVVd4WLzgkXRHxl-I27yWtz?usp=sharing)
 
 
-### Aidatatang_200zh
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1MmY5bBxwvKLNT4A2DJnwiqRXhdchUqPN?usp=sharing)
 
-We provide one model for this recipe: [Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][Aidatatang_200zh_pruned_transducer_stateless2].
+#### [Transducer (pruned_transducer_stateless)](https://github.com/k2-fsa/icefall/tree/master/egs/tedlium3/ASR/pruned_transducer_stateless)
 
-#### Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss
+|                                      |  dev  |  test  |
+|--------------------------------------|-------|--------|
+| modified_beam_search (`beam_size=4`) |  6.77 |  6.14  |
+
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1je_1zGrOkGVVd4WLzgkXRHxl-I27yWtz?usp=sharing)
+
+
+### [Aidatatang_200zh][aidatatang_200zh]
+
+#### [Transducer (pruned_transducer_stateless2)](https://github.com/k2-fsa/icefall/tree/master/egs/aidatatang_200zh/ASR/pruned_transducer_stateless2)
 
 |                      |  Dev  | Test  |
 |----------------------|-------|-------|
-|    greedy search     | 5.53  | 6.59  |
+|    greedy_search     | 5.53  | 6.59  |
-|   fast beam search   | 5.30  | 6.34  |
+|   fast_beam_search   | 5.30  | 6.34  |
-| modified beam search | 5.27  | 6.33  |
+| modified_beam_search | 5.27  | 6.33  |
 
-We provide a Colab notebook to run a pre-trained Pruned Transducer Stateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1wNSnSj3T5oOctbh5IGCa393gKOoQw2GH?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1wNSnSj3T5oOctbh5IGCa393gKOoQw2GH?usp=sharing)
 
 
-### WenetSpeech
+### [WenetSpeech][wenetspeech]
 
-We provide some models for this recipe: [Pruned stateless RNN-T_2: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][WenetSpeech_pruned_transducer_stateless2] and [Pruned stateless RNN-T_5: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][WenetSpeech_pruned_transducer_stateless5].
+#### [Transducer (pruned_transducer_stateless2)](https://github.com/k2-fsa/icefall/tree/master/egs/wenetspeech/ASR/pruned_transducer_stateless2)
-
-#### Pruned stateless RNN-T_2: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss (trained with L subset, offline ASR)
 
 |                      |  Dev  | Test-Net | Test-Meeting |
 |----------------------|-------|----------|--------------|
-|    greedy search     | 7.80  |  8.75    |  13.49       |
+|    greedy_search     | 7.80  |  8.75    |  13.49       |
-| modified beam search| 7.76  |  8.71    |  13.41       |
+|   fast_beam_search   | 7.94  |  8.74    |  13.80       |
-|   fast beam search   | 7.94  |  8.74    |  13.80       |
+| modified_beam_search | 7.76  |  8.71    |  13.41       |
+
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1EV4e1CHa1GZgEF-bZgizqI9RyFFehIiN?usp=sharing)
+
+#### [Transducer **Streaming** (pruned_transducer_stateless5) ](https://github.com/k2-fsa/icefall/tree/master/egs/wenetspeech/ASR/pruned_transducer_stateless5)
 
-#### Pruned stateless RNN-T_5: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss (trained with L subset)
-**Streaming**:
 |                      |  Dev  | Test-Net | Test-Meeting |
 |----------------------|-------|----------|--------------|
 | greedy_search | 8.78 | 10.12 | 16.16 |
-| modified_beam_search | 8.53| 9.95 | 15.81 |
 | fast_beam_search| 9.01 | 10.47 | 16.28 |
+| modified_beam_search | 8.53| 9.95 | 15.81 |
 
-We provide a Colab notebook to run a pre-trained Pruned Transducer Stateless2 model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1EV4e1CHa1GZgEF-bZgizqI9RyFFehIiN?usp=sharing)
 
-### Alimeeting
+### [Alimeeting][alimeeting]
 
-We provide one model for this recipe: [Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][Alimeeting_pruned_transducer_stateless2].
+#### [Transducer (pruned_transducer_stateless2)](https://github.com/k2-fsa/icefall/tree/master/egs/alimeeting/ASR/pruned_transducer_stateless2)
-
-#### Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss (trained with far subset)
 
 |                      |  Eval  | Test-Net |
 |----------------------|--------|----------|
-|    greedy search     | 31.77  |  34.66   |
+|    greedy_search     | 31.77  |  34.66   |
-|   fast beam search   | 31.39  |  33.02   |
+|   fast_beam_search   | 31.39  |  33.02   |
-| modified beam search | 30.38  |  34.25   |
+| modified_beam_search | 30.38  |  34.25   |
 
-We provide a Colab notebook to run a pre-trained Pruned Transducer Stateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1tKr3f0mL17uO_ljdHGKtR7HOmthYHwJG?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1tKr3f0mL17uO_ljdHGKtR7HOmthYHwJG?usp=sharing)
 
 
-### TAL_CSASR
+### [TAL_CSASR][tal_csasr]
 
-We provide one model for this recipe: [Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss][TAL_CSASR_pruned_transducer_stateless5].
 
-#### Pruned stateless RNN-T: Conformer encoder + Embedding decoder + k2 pruned RNN-T loss
+#### [Transducer (pruned_transducer_stateless5)](https://github.com/k2-fsa/icefall/tree/master/egs/tal_csasr/ASR/pruned_transducer_stateless5)
 
 The best results for Chinese CER(%) and English WER(%) respectively (zh: Chinese, en: English):
 |decoding-method | dev | dev_zh | dev_en | test | test_zh | test_en |
 |--|--|--|--|--|--|--|
 |greedy_search| 7.30 | 6.48 | 19.19 |7.39| 6.66 | 19.13|
-|modified_beam_search| 7.15 | 6.35 | 18.95 | 7.22| 6.50 | 18.70 |
 |fast_beam_search| 7.18 | 6.39| 18.90 |  7.27| 6.55 | 18.77|
+|modified_beam_search| 7.15 | 6.35 | 18.95 | 7.22| 6.50 | 18.70 |
 
-We provide a Colab notebook to run a pre-trained Pruned Transducer Stateless model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1DmIx-NloI1CMU5GdZrlse7TRu4y3Dpf8?usp=sharing)
+We provide a Colab notebook to test the pre-trained model: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1DmIx-NloI1CMU5GdZrlse7TRu4y3Dpf8?usp=sharing)
 
-## Deployment with C++
+## TTS: Text-to-Speech
 
-Once you have trained a model in icefall, you may want to deploy it with C++,
+### Supported Datasets
-without Python dependencies.
 
-Please refer to the documentation
+  - [LJSpeech][ljspeech]
-<https://icefall.readthedocs.io/en/latest/recipes/Non-streaming-ASR/librispeech/conformer_ctc.html#deployment-with-c>
+  - [VCTK][vctk]
+
+### Supported Models
+
+  - [VITS](https://arxiv.org/abs/2106.06103)
+
+# Deployment with C++
+
+Once you have trained a model in icefall, you may want to deploy it with C++ without Python dependencies.
+
+Please refer to the [document](https://icefall.readthedocs.io/en/latest/recipes/Non-streaming-ASR/librispeech/conformer_ctc.html#deployment-with-c)
 for how to do this.
 
 We also provide a Colab notebook, showing you how to run a torch scripted model in [k2][k2] with C++.
 Please see: [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1BIGLWzS36isskMXHKcqC9ysN6pspYXs_?usp=sharing)
 
 
-[LibriSpeech_tdnn_lstm_ctc]: egs/librispeech/ASR/tdnn_lstm_ctc
-[LibriSpeech_conformer_ctc]: egs/librispeech/ASR/conformer_ctc
-[LibriSpeech_transducer]: egs/librispeech/ASR/transducer
-[LibriSpeech_transducer_stateless]: egs/librispeech/ASR/transducer_stateless
-[LibriSpeech_zipformer]: egs/librispeech/ASR/zipformer
-[Aishell_tdnn_lstm_ctc]: egs/aishell/ASR/tdnn_lstm_ctc
-[Aishell_conformer_ctc]: egs/aishell/ASR/conformer_ctc
-[Aishell_pruned_transducer_stateless7]: egs/aishell/ASR/pruned_transducer_stateless7_bbpe
-[Aishell2_pruned_transducer_stateless5]: egs/aishell2/ASR/pruned_transducer_stateless5
-[Aishell4_pruned_transducer_stateless5]: egs/aishell4/ASR/pruned_transducer_stateless5
-[TIMIT_tdnn_lstm_ctc]: egs/timit/ASR/tdnn_lstm_ctc
-[TIMIT_tdnn_ligru_ctc]: egs/timit/ASR/tdnn_ligru_ctc
-[TED-LIUM3_transducer_stateless]: egs/tedlium3/ASR/transducer_stateless
-[TED-LIUM3_pruned_transducer_stateless]: egs/tedlium3/ASR/pruned_transducer_stateless
-[GigaSpeech_conformer_ctc]: egs/gigaspeech/ASR/conformer_ctc
-[GigaSpeech_pruned_transducer_stateless2]: egs/gigaspeech/ASR/pruned_transducer_stateless2
-[GigaSpeech_zipformer]: egs/gigaspeech/ASR/zipformer
-[Aidatatang_200zh_pruned_transducer_stateless2]: egs/aidatatang_200zh/ASR/pruned_transducer_stateless2
-[WenetSpeech_pruned_transducer_stateless2]: egs/wenetspeech/ASR/pruned_transducer_stateless2
-[WenetSpeech_pruned_transducer_stateless5]: egs/wenetspeech/ASR/pruned_transducer_stateless5
-[Alimeeting_pruned_transducer_stateless2]: egs/alimeeting/ASR/pruned_transducer_stateless2
-[TAL_CSASR_pruned_transducer_stateless5]: egs/tal_csasr/ASR/pruned_transducer_stateless5
 [yesno]: egs/yesno/ASR
 [librispeech]: egs/librispeech/ASR
 [aishell]: egs/aishell/ASR
@@ -411,3 +365,15 @@ Please see: [![Open In Colab](https://colab.research.google.com/assets/colab-bad
 [ami]: egs/ami
 [swbd]: egs/swbd/ASR
 [k2]: https://github.com/k2-fsa/k2
+[commonvoice]: egs/commonvoice/ASR
+[csj]: egs/csj/ASR
+[libricss]: egs/libricss/SURT
+[libriheavy]: egs/libriheavy/ASR
+[mgb2]: egs/mgb2/ASR
+[peoplespeech]: egs/peoplespeech/ASR
+[spgispeech]: egs/spgispeech/ASR
+[voxpopuli]: egs/voxpopuli/ASR
+[xbmu-amdo31]: egs/xbmu-amdo31/ASR
+
+[vctk]: egs/vctk/TTS
+[ljspeech]: egs/ljspeech/TTS

docker/torch1.12.1-cuda11.3.dockerfile

@@ -18,7 +18,7 @@ RUN apt-get update && \
     apt-get install -y --no-install-recommends \
         curl \
         vim \
-    		libssl-dev \
+    	libssl-dev \
         autoconf \
         automake \
         bzip2 \
@@ -44,7 +44,6 @@ RUN pip install --no-cache-dir \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
-      \
       kaldi_native_io \
       kaldialign \
       kaldifst \

docker/torch1.13.0-cuda11.6.dockerfile

@@ -18,7 +18,7 @@ RUN apt-get update && \
     apt-get install -y --no-install-recommends \
         curl \
         vim \
-    		libssl-dev \
+    	libssl-dev \
         autoconf \
         automake \
         bzip2 \
@@ -44,7 +44,6 @@ RUN pip install --no-cache-dir \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
-      \
       kaldi_native_io \
       kaldialign \
       kaldifst \

docker/torch1.9.0-cuda10.2.dockerfile

@@ -25,7 +25,7 @@ RUN apt-get update && \
     apt-get install -y --no-install-recommends \
         curl \
         vim \
-    		libssl-dev \
+    	libssl-dev \
         autoconf \
         automake \
         bzip2 \
@@ -58,7 +58,6 @@ RUN pip uninstall -y tqdm && \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
-      \
       kaldi_native_io \
       kaldialign \
       kaldifst \

docker/torch2.0.0-cuda11.7.dockerfile

@@ -18,7 +18,7 @@ RUN apt-get update && \
     apt-get install -y --no-install-recommends \
         curl \
         vim \
-    		libssl-dev \
+    	libssl-dev \
         autoconf \
         automake \
         bzip2 \
@@ -44,7 +44,6 @@ RUN pip install --no-cache-dir \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
-      \
       kaldi_native_io \
       kaldialign \
       kaldifst \

docker/torch2.1.0-cuda11.8.dockerfile

@@ -18,7 +18,7 @@ RUN apt-get update && \
     apt-get install -y --no-install-recommends \
         curl \
         vim \
-    		libssl-dev \
+    	libssl-dev \
         autoconf \
         automake \
         bzip2 \
@@ -44,7 +44,6 @@ RUN pip install --no-cache-dir \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
-      \
       kaldi_native_io \
       kaldialign \
       kaldifst \

docker/torch2.1.0-cuda12.1.dockerfile

@@ -18,7 +18,7 @@ RUN apt-get update && \
     apt-get install -y --no-install-recommends \
         curl \
         vim \
-    		libssl-dev \
+    	libssl-dev \
         autoconf \
         automake \
         bzip2 \
@@ -44,7 +44,6 @@ RUN pip install --no-cache-dir \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
-      \
       kaldi_native_io \
       kaldialign \
       kaldifst \

docs/source/docker/intro.rst

@@ -20,7 +20,11 @@ We describe the following items in this section:
 View available tags
 ===================
 
-You can use the following command to view available tags:
+CUDA-enabled docker images
+--------------------------
+
+You can use the following command to view available tags for CUDA-enabled
+docker images:
 
 .. code-block:: bash
 
@@ -43,8 +47,25 @@ which will give you something like below:
 
 Please select an appropriate combination of `torch`_ and  CUDA.
 
-Download a docker image
+CPU-only docker images
-=======================
+----------------------
+
+To view CPU-only docker images, please visit `<https://github.com/k2-fsa/icefall/pkgs/container/icefall>`_
+for available tags.
+
+You can select different combinations of ``Python`` and ``torch``. For instance,
+to select ``Python 3.8`` and ``torch 2.1.2``, you can use the following tag
+
+.. code-block:: bash
+
+    cpu-py3.8-torch2.1.2-v1.1
+
+where ``v1.1`` is the current version of the docker image. You may see
+``ghcr.io/k2-fsa/icefall:cpu-py3.8-torch2.1.2-v1.2`` or some other versions.
+We recommend that you always use the latest version.
+
+Download a docker image (CUDA)
+==============================
 
 Suppose that you select the tag ``torch1.13.0-cuda11.6``, you can use
 the following command to download it:
@@ -53,6 +74,16 @@ the following command to download it:
 
    sudo docker image pull k2fsa/icefall:torch1.13.0-cuda11.6
 
+Download a docker image (CPU)
+==============================
+
+Suppose that you select the tag ``cpu-py3.8-torch2.1.2-v1.1``, you can use
+the following command to download it:
+
+.. code-block:: bash
+
+  sudo docker pull ghcr.io/k2-fsa/icefall:cpu-py3.8-torch2.1.2-v1.1
+
 Run a docker image with GPU
 ===========================
 
@@ -65,7 +96,7 @@ Run a docker image with CPU
 
 .. code-block:: bash
 
-  sudo docker run --rm -it k2fsa/icefall:torch1.13.0-cuda11.6 /bin/bash
+  sudo docker run --rm -it ghcr.io/k2-fsa/icefall:cpu-py3.8-torch2.1.2-v1.1 /bin/bash
 
 Run yesno within a docker container
 ===================================
@@ -74,8 +105,13 @@ After starting the container, the following interface is presented:
 
 .. code-block:: bash
 
+  # GPU-enabled docker
   root@60c947eac59c:/workspace/icefall#
 
+  # CPU-only docker
+  root@60c947eac59c:# mkdir /workspace; git clone https://github.com/k2-fsa/icefall
+  root@60c947eac59c:# export PYTHONPATH=/workspace/icefall:$PYTHONPATH
+
 It shows the current user is ``root`` and the current working directory
 is ``/workspace/icefall``.
 
@@ -107,7 +143,7 @@ to switch to the ``yesno`` recipe and run
 
 .. hint::
 
-   If you are running without GPU, it may report the following error:
+   If you are running without GPU with a GPU-enabled docker, it may report the following error:
 
     .. code-block:: bash
 

docs/source/for-dummies/environment-setup.rst

@@ -66,13 +66,13 @@ to install dependencies of `icefall`_:
 
    pip install torch==2.0.0+cpu torchaudio==2.0.0+cpu -f https://download.pytorch.org/whl/torch_stable.html
 
-   # If you are using macOS or Windows, please use the following command to install torch and torchaudio
+   # If you are using macOS, please use the following command to install torch and torchaudio
    # pip install torch==2.0.0 torchaudio==2.0.0 -f https://download.pytorch.org/whl/torch_stable.html
 
    # Now install k2
    # Please refer to https://k2-fsa.github.io/k2/installation/from_wheels.html#linux-cpu-example
 
-   pip install k2==1.24.3.dev20230726+cpu.torch2.0.0 -f https://k2-fsa.github.io/k2/cpu.html
+   pip install k2==1.24.4.dev20231220+cpu.torch2.0.0 -f https://k2-fsa.github.io/k2/cpu.html
 
    # Install the latest version of lhotse
 

docs/source/for-dummies/model-export.rst

@@ -85,7 +85,7 @@ We can also use it to decode files with the following command:
   # Please refer to https://csukuangfj.github.io/kaldifeat/installation/from_wheels.html
   # for how to install kaldifeat
 
-  pip install kaldifeat==1.25.0.dev20230726+cpu.torch2.0.0 -f https://csukuangfj.github.io/kaldifeat/cpu.html
+  pip install kaldifeat==1.25.3.dev20231221+cpu.torch2.0.0 -f https://csukuangfj.github.io/kaldifeat/cpu.html
 
   ./tdnn/pretrained.py \
     --checkpoint ./tdnn/exp/pretrained.pt \
@@ -162,7 +162,7 @@ To use ``tdnn/exp/cpu_jit.pt`` with `icefall`_ to decode files, we can use:
   # Please refer to https://csukuangfj.github.io/kaldifeat/installation/from_wheels.html
   # for how to install kaldifeat
 
-  pip install kaldifeat==1.25.0.dev20230726+cpu.torch2.0.0 -f https://csukuangfj.github.io/kaldifeat/cpu.html
+  pip install kaldifeat==1.25.3.dev20231221+cpu.torch2.0.0 -f https://csukuangfj.github.io/kaldifeat/cpu.html
 
 
   ./tdnn/jit_pretrained.py \
@@ -249,7 +249,7 @@ To use the generated ONNX model files for decoding with `onnxruntime`_, we can u
   # Please refer to https://csukuangfj.github.io/kaldifeat/installation/from_wheels.html
   # for how to install kaldifeat
 
-  pip install kaldifeat==1.25.0.dev20230726+cpu.torch2.0.0 -f https://csukuangfj.github.io/kaldifeat/cpu.html
+  pip install kaldifeat==1.25.3.dev20231221+cpu.torch2.0.0 -f https://csukuangfj.github.io/kaldifeat/cpu.html
 
   ./tdnn/onnx_pretrained.py \
     --nn-model ./tdnn/exp/model-epoch-14-avg-2.onnx \

docs/source/recipes/RNN-LM/librispeech/lm-training.rst

@@ -77,7 +77,7 @@ The next step is to train the RNNLM model. The training command is as follows:
         --use-fp16 0 \
         --tie-weights 1 \
         --embedding-dim 2048 \
-        --hidden_dim 2048 \
+        --hidden-dim 2048 \
         --num-layers 3 \
         --batch-size 300 \
         --lm-data rnn_lm/data/lang_bpe_500/sorted_lm_data.pt \
@@ -93,12 +93,3 @@ The next step is to train the RNNLM model. The training command is as follows:
 .. note::
 
     The training of RNNLM can take a long time (usually a couple of days).
-
-
-
-
-
-
-
-
-

docs/source/recipes/TTS/index.rst

@@ -0,0 +1,8 @@
+TTS
+======
+
+.. toctree::
+   :maxdepth: 2
+
+   ljspeech/vits
+   vctk/vits

docs/source/recipes/TTS/ljspeech/vits.rst

@@ -0,0 +1,123 @@
+VITS
+===============
+
+This tutorial shows you how to train an VITS model
+with the `LJSpeech <https://keithito.com/LJ-Speech-Dataset/>`_ dataset.
+
+.. note::
+  
+   TTS related recipes require packages in ``requirements-tts.txt``.
+
+.. note::
+
+   The VITS paper: `Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech <https://arxiv.org/pdf/2106.06103.pdf>`_
+
+
+Data preparation
+----------------
+
+.. code-block:: bash
+
+  $ cd egs/ljspeech/TTS
+  $ ./prepare.sh
+
+To run stage 1 to stage 5, use
+
+.. code-block:: bash
+
+  $ ./prepare.sh --stage 1 --stop_stage 5
+
+
+Build Monotonic Alignment Search
+--------------------------------
+
+.. code-block:: bash
+
+  $ ./prepare.sh --stage -1 --stop_stage -1
+
+or
+
+.. code-block:: bash
+
+  $ cd vits/monotonic_align
+  $ python setup.py build_ext --inplace
+  $ cd ../../
+
+
+Training
+--------
+
+.. code-block:: bash
+
+  $ export CUDA_VISIBLE_DEVICES="0,1,2,3"
+  $ ./vits/train.py \
+      --world-size 4 \
+      --num-epochs 1000 \
+      --start-epoch 1 \
+      --use-fp16 1 \
+      --exp-dir vits/exp \
+      --tokens data/tokens.txt
+      --max-duration 500
+
+.. note::
+
+    You can adjust the hyper-parameters to control the size of the VITS model and
+    the training configurations. For more details, please run ``./vits/train.py --help``.
+
+.. note::
+
+    The training can take a long time (usually a couple of days).
+
+Training logs, checkpoints and tensorboard logs are saved in ``vits/exp``.
+
+
+Inference
+---------
+
+The inference part uses checkpoints saved by the training part, so you have to run the
+training part first. It will save the ground-truth and generated wavs to the directory
+``vits/exp/infer/epoch-*/wav``, e.g., ``vits/exp/infer/epoch-1000/wav``.
+
+.. code-block:: bash
+
+  $ export CUDA_VISIBLE_DEVICES="0"
+  $ ./vits/infer.py \
+      --epoch 1000 \
+      --exp-dir vits/exp \
+      --tokens data/tokens.txt \
+      --max-duration 500
+
+.. note::
+
+    For more details, please run ``./vits/infer.py --help``.
+
+
+Export models
+-------------
+
+Currently we only support ONNX model exporting. It will generate two files in the given ``exp-dir``:
+``vits-epoch-*.onnx`` and ``vits-epoch-*.int8.onnx``.
+
+.. code-block:: bash
+
+  $ ./vits/export-onnx.py \
+      --epoch 1000 \
+      --exp-dir vits/exp \
+      --tokens data/tokens.txt
+
+You can test the exported ONNX model with:
+
+.. code-block:: bash
+
+  $ ./vits/test_onnx.py \
+      --model-filename vits/exp/vits-epoch-1000.onnx \
+      --tokens data/tokens.txt
+
+
+Download pretrained models
+--------------------------
+
+If you don't want to train from scratch, you can download the pretrained models
+by visiting the following link:
+
+  - `<https://huggingface.co/Zengwei/icefall-tts-ljspeech-vits-2023-11-29>`_

docs/source/recipes/TTS/vctk/vits.rst

@@ -0,0 +1,125 @@
+VITS
+===============
+
+This tutorial shows you how to train an VITS model
+with the `VCTK <https://datashare.ed.ac.uk/handle/10283/3443>`_ dataset.
+
+.. note::
+  
+   TTS related recipes require packages in ``requirements-tts.txt``.
+
+.. note::
+
+   The VITS paper: `Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech <https://arxiv.org/pdf/2106.06103.pdf>`_
+
+
+Data preparation
+----------------
+
+.. code-block:: bash
+
+  $ cd egs/vctk/TTS
+  $ ./prepare.sh
+
+To run stage 1 to stage 6, use
+
+.. code-block:: bash
+
+  $ ./prepare.sh --stage 1 --stop_stage 6
+
+
+Build Monotonic Alignment Search
+--------------------------------
+
+To build the monotonic alignment search, use the following commands:
+
+.. code-block:: bash
+
+  $ ./prepare.sh --stage -1 --stop_stage -1
+
+or
+
+.. code-block:: bash
+
+  $ cd vits/monotonic_align
+  $ python setup.py build_ext --inplace
+  $ cd ../../
+
+
+Training
+--------
+
+.. code-block:: bash
+
+  $ export CUDA_VISIBLE_DEVICES="0,1,2,3"
+  $ ./vits/train.py \
+      --world-size 4 \
+      --num-epochs 1000 \
+      --start-epoch 1 \
+      --use-fp16 1 \
+      --exp-dir vits/exp \
+      --tokens data/tokens.txt
+      --max-duration 350
+
+.. note::
+
+    You can adjust the hyper-parameters to control the size of the VITS model and
+    the training configurations. For more details, please run ``./vits/train.py --help``.
+
+.. note::
+
+    The training can take a long time (usually a couple of days).
+
+Training logs, checkpoints and tensorboard logs are saved in ``vits/exp``.
+
+
+Inference
+---------
+
+The inference part uses checkpoints saved by the training part, so you have to run the
+training part first. It will save the ground-truth and generated wavs to the directory
+``vits/exp/infer/epoch-*/wav``, e.g., ``vits/exp/infer/epoch-1000/wav``.
+
+.. code-block:: bash
+
+  $ export CUDA_VISIBLE_DEVICES="0"
+  $ ./vits/infer.py \
+      --epoch 1000 \
+      --exp-dir vits/exp \
+      --tokens data/tokens.txt \
+      --max-duration 500
+
+.. note::
+
+    For more details, please run ``./vits/infer.py --help``.
+
+
+Export models
+-------------
+
+Currently we only support ONNX model exporting. It will generate two files in the given ``exp-dir``:
+``vits-epoch-*.onnx`` and ``vits-epoch-*.int8.onnx``.
+
+.. code-block:: bash
+
+  $ ./vits/export-onnx.py \
+      --epoch 1000 \
+      --exp-dir vits/exp \
+      --tokens data/tokens.txt
+
+You can test the exported ONNX model with:
+
+.. code-block:: bash
+
+  $ ./vits/test_onnx.py \
+      --model-filename vits/exp/vits-epoch-1000.onnx \
+      --tokens data/tokens.txt
+
+
+Download pretrained models
+--------------------------
+
+If you don't want to train from scratch, you can download the pretrained models
+by visiting the following link:
+
+  - `<https://huggingface.co/zrjin/icefall-tts-vctk-vits-2023-12-05>`_

docs/source/recipes/index.rst

@@ -2,7 +2,7 @@ Recipes
 =======
 
 This page contains various recipes in ``icefall``.
-Currently, only speech recognition recipes are provided.
+Currently, we provide recipes for speech recognition, language model, and speech synthesis.
 
 We may add recipes for other tasks as well in the future.
 
@@ -16,3 +16,4 @@ We may add recipes for other tasks as well in the future.
    Non-streaming-ASR/index
    Streaming-ASR/index
    RNN-LM/index
+   TTS/index

egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/asr_datamodule.py

@@ -288,8 +288,9 @@ class Aidatatang_200zhAsrDataModule:
                 max_duration=self.args.max_duration,
                 shuffle=self.args.shuffle,
                 num_buckets=self.args.num_buckets,
+                buffer_size=self.args.num_buckets * 2000,
+                shuffle_buffer_size=self.args.num_buckets * 5000,
                 drop_last=True,
-                buffer_size=50000,
             )
         else:
             logging.info("Using SimpleCutSampler.")

egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/export.py

@@ -1,3 +1,4 @@
+#!/usr/bin/env python3
 # Copyright 2021 Xiaomi Corporation (Author: Fangjun Kuang)
 #
 # See ../../../../LICENSE for clarification regarding multiple authors
@@ -20,7 +21,7 @@
 Usage:
 ./pruned_transducer_stateless2/export.py \
   --exp-dir ./pruned_transducer_stateless2/exp \
-  --lang-dir data/lang_char \
+  --tokens data/lang_char/tokens.txt \
   --epoch 29 \
   --avg 19
 
@@ -45,12 +46,13 @@ import argparse
 import logging
 from pathlib import Path
 
+import k2
 import torch
+from scaling_converter import convert_scaled_to_non_scaled
 from train import get_params, get_transducer_model
 
 from icefall.checkpoint import average_checkpoints, load_checkpoint
-from icefall.lexicon import Lexicon
+from icefall.utils import num_tokens, str2bool
-from icefall.utils import str2bool
 
 
 def get_parser():
@@ -85,10 +87,10 @@ def get_parser():
     )
 
     parser.add_argument(
-        "--lang-dir",
+        "--tokens",
         type=str,
-        default="data/lang_char",
+        default="data/lang_char/tokens.txt",
-        help="The lang dir",
+        help="Path to the tokens.txt.",
     )
 
     parser.add_argument(
@@ -122,10 +124,14 @@ def main():
 
     logging.info(f"device: {device}")
 
-    lexicon = Lexicon(params.lang_dir)
+    # Load tokens.txt here
+    token_table = k2.SymbolTable.from_file(params.tokens)
 
-    params.blank_id = 0
+    # Load id of the <blk> token and the vocab size
-    params.vocab_size = max(lexicon.tokens) + 1
+    # <blk> is defined in local/train_bpe_model.py
+    params.blank_id = token_table["<blk>"]
+    params.unk_id = token_table["<unk>"]
+    params.vocab_size = num_tokens(token_table) + 1  # +1 for <blk>
 
     logging.info(params)
 
@@ -152,6 +158,7 @@ def main():
     model.eval()
 
     if params.jit:
+        convert_scaled_to_non_scaled(model, inplace=True)
         # We won't use the forward() method of the model in C++, so just ignore
         # it here.
         # Otherwise, one of its arguments is a ragged tensor and is not

egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/lstmp.py

@@ -0,0 +1 @@
+../../../librispeech/ASR/lstm_transducer_stateless2/lstmp.py

egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/pretrained.py

@@ -242,6 +242,7 @@ def main():
     opts.frame_opts.snip_edges = False
     opts.frame_opts.samp_freq = params.sample_rate
     opts.mel_opts.num_bins = params.feature_dim
+    opts.mel_opts.high_freq = -400
 
     fbank = kaldifeat.Fbank(opts)
 

egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/scaling_converter.py

@@ -0,0 +1 @@
+../../../librispeech/ASR/pruned_transducer_stateless3/scaling_converter.py

egs/aishell/ASR/README.md

@@ -24,3 +24,10 @@ The following table lists the differences among them.
 The decoder in `transducer_stateless` is modified from the paper
 [Rnn-Transducer with Stateless Prediction Network](https://ieeexplore.ieee.org/document/9054419/).
 We place an additional Conv1d layer right after the input embedding layer.
+
+# Whisper
+
+Recipe to finetune large pretrained models
+|                                    | Encoder   | Decoder            | Comment                                                                           |
+|------------------------------------|-----------|--------------------|-----------------------------------------------------------------------------------|
+| `whisper`             | Transformer | Transformer | support fine-tuning using deepspeed

egs/aishell/ASR/RESULTS.md

@@ -1,10 +1,120 @@
 ## Results
 
+### Aishell training results (Fine-tuning Pretrained Models)
+#### Whisper
+[./whisper](./whisper)
+##### fine-tuning results on Aishell test set on whisper medium, large-v2, large-v3
+
+|                        | test (before fine-tuning) | test (after fine-tuning)  | comment                                 |
+|------------------------|------|------|-----------------------------------------|
+| medium         | 7.23 | 3.27 | --epoch 10 --avg 4,  ddp                         |
+| large-v2       | 6.56 | 2.47 | --epoch 10 --avg 6,  deepspeed zero stage1       |
+| large-v3       | 6.06 | 2.84 | --epoch 5 --avg 3,   deepspeed zero stage1       |
+
+Command for training is:
+```bash
+pip install -r whisper/requirements.txt
+
+./prepare.sh --stage 30 --stop_stage 30
+
+#fine-tuning with deepspeed zero stage 1
+torchrun --nproc-per-node 8 ./whisper/train.py \
+  --max-duration 200 \
+  --exp-dir whisper/exp_large_v2 \
+  --model-name large-v2 \
+  --deepspeed \
+  --deepspeed_config ./whisper/ds_config_zero1.json
+
+# fine-tuning with ddp
+torchrun --nproc-per-node 8 ./whisper/train.py \
+  --max-duration 200 \
+  --exp-dir whisper/exp_medium \
+  --base-lr 1e-5 \
+  --model-name medium
+```
+
+Command for decoding using fine-tuned models:
+```bash
+git lfs install
+git clone https://huggingface.co/yuekai/icefall_asr_aishell_whisper
+ln -s icefall_asr_aishell_whisper/exp_large_v2/epoch-10-avg6.pt whisper/exp_large_v2/epoch-999.pt
+
+python3 ./whisper/decode.py \
+  --exp-dir whisper/exp_large_v2 \
+  --model-name large-v2 \
+  --epoch 999 --avg 1 \
+  --beam-size 10 --max-duration 50
+```
+Command for decoding using pretrained models (before fine-tuning):
+```bash
+python3 ./whisper/decode.py \
+  --exp-dir whisper/exp_large_v2 \
+  --model-name large-v2 \
+  --epoch -1 --avg 1 \
+  --remove-whisper-encoder-input-length-restriction False \
+  --beam-size 10 --max-duration 50
+```
+Fine-tuned models, training logs, decoding logs, tensorboard and decoding results
+are available at
+<https://huggingface.co/yuekai/icefall_asr_aishell_whisper>
+
 ### Aishell training result (Stateless Transducer)
 
+#### Zipformer (Byte-level BPE)
+
+[./zipformer](./zipformer/)
+
+It's reworked Zipformer with Pruned RNNT loss, trained with Byte-level BPE, `vocab_size` set to 500.
+
+##### normal-scaled model, number of model parameters: 65549011, i.e., 65.55 M
+
+|                        | test | dev  | comment                                 |
+|------------------------|------|------|-----------------------------------------|
+| greedy search          | 4.54 | 4.31 | --epoch 40 --avg 10                     |
+| modified beam search   | 4.37 | 4.11 | --epoch 40 --avg 10                     |
+| fast beam search       | 4.43 | 4.17 | --epoch 40 --avg 10                     |
+
+```bash
+./prepare.sh 
+
+export CUDA_VISIBLE_DEVICES="0,1"
+
+./zipformer/train_bbpe.py \
+  --world-size 2 \
+  --num-epochs 40 \
+  --start-epoch 1 \
+  --use-fp16 1 \
+  --context-size 2 \
+  --enable-musan 0 \
+  --exp-dir zipformer/exp_bbpe \
+  --max-duration 1000 \
+  --enable-musan 0 \
+  --base-lr 0.045 \
+  --lr-batches 7500 \
+  --lr-epochs 10 \
+  --spec-aug-time-warp-factor 20
+```
+
+Command for decoding is:
+```bash
+for m in greedy_search modified_beam_search fast_beam_search ; do
+  ./zipformer/decode_bbpe.py \
+    --epoch 40 \
+    --avg 10 \
+    --exp-dir ./zipformer_bbpe/exp \
+    --bpe-model data/lang_bbpe_500/bbpe.model \
+    --context-size 2 \
+    --decoding-method $m
+done
+```
+Pretrained models, training logs, decoding logs, tensorboard and decoding results
+are available at
+<https://huggingface.co/zrjin/icefall-asr-aishell-zipformer-bbpe-2024-01-16>
+
+
 #### Zipformer (Non-streaming)
 
-[./zipformer](./zipformer)
+[./zipformer](./zipformer/)
 
 It's reworked Zipformer with Pruned RNNT loss.
 **Caution**: It uses `--context-size=1`.
@@ -260,7 +370,7 @@ done
 Pretrained models, training logs, decoding logs, and decoding results
 are available at
 <https://huggingface.co/marcoyang/icefall-asr-aishell-zipformer-pruned-transducer-stateless7-2023-03-21>
-#### Pruned transducer stateless 7 (zipformer)
+#### Pruned transducer stateless 7 (Byte-level BPE)
 
 See <https://github.com/k2-fsa/icefall/pull/986>
 
@@ -703,7 +813,6 @@ python3 ./transducer_stateless/decode.py \
        --max-sym-per-frame 3
 ```
 
-### Aishell training results (Transducer-stateless)
 #### 2022-02-18
 (Pingfeng Luo) : The tensorboard log for training is available at <https://tensorboard.dev/experiment/k3QL6QMhRbCwCKYKM9po9w/>
 And pretrained model is available at <https://huggingface.co/pfluo/icefall-aishell-transducer-stateless-char-2021-12-29>

egs/aishell/ASR/conformer_ctc/pretrained.py

@@ -261,6 +261,7 @@ def main():
     opts.frame_opts.snip_edges = False
     opts.frame_opts.samp_freq = params.sample_rate
     opts.mel_opts.num_bins = params.feature_dim
+    opts.mel_opts.high_freq = -400
 
     fbank = kaldifeat.Fbank(opts)
 

egs/aishell/ASR/local/compute_fbank_aishell.py

@@ -29,7 +29,14 @@ import os
 from pathlib import Path
 
 import torch
-from lhotse import CutSet, Fbank, FbankConfig, LilcomChunkyWriter
+from lhotse import (
+    CutSet,
+    Fbank,
+    FbankConfig,
+    LilcomChunkyWriter,
+    WhisperFbank,
+    WhisperFbankConfig,
+)
 from lhotse.recipes.utils import read_manifests_if_cached
 
 from icefall.utils import get_executor, str2bool
@@ -42,9 +49,14 @@ torch.set_num_threads(1)
 torch.set_num_interop_threads(1)
 
 
-def compute_fbank_aishell(num_mel_bins: int = 80, perturb_speed: bool = False):
+def compute_fbank_aishell(
+    num_mel_bins: int = 80,
+    perturb_speed: bool = False,
+    whisper_fbank: bool = False,
+    output_dir: str = "data/fbank",
+):
     src_dir = Path("data/manifests")
-    output_dir = Path("data/fbank")
+    output_dir = Path(output_dir)
     num_jobs = min(15, os.cpu_count())
 
     dataset_parts = (
@@ -68,8 +80,12 @@ def compute_fbank_aishell(num_mel_bins: int = 80, perturb_speed: bool = False):
         list(manifests.keys()),
         dataset_parts,
     )
-
+    if whisper_fbank:
-    extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
+        extractor = WhisperFbank(
+            WhisperFbankConfig(num_filters=num_mel_bins, device="cuda")
+        )
+    else:
+        extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.
         for partition, m in manifests.items():
@@ -82,7 +98,7 @@ def compute_fbank_aishell(num_mel_bins: int = 80, perturb_speed: bool = False):
                 supervisions=m["supervisions"],
             )
             if "train" in partition and perturb_speed:
-                logging.info(f"Doing speed perturb")
+                logging.info("Doing speed perturb")
                 cut_set = (
                     cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
                 )
@@ -111,6 +127,18 @@ def get_args():
         default=False,
         help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
     )
+    parser.add_argument(
+        "--whisper-fbank",
+        type=str2bool,
+        default=False,
+        help="Use WhisperFbank instead of Fbank. Default: False.",
+    )
+    parser.add_argument(
+        "--output-dir",
+        type=str,
+        default="data/fbank",
+        help="Output directory. Default: data/fbank.",
+    )
     return parser.parse_args()
 
 
@@ -121,5 +149,8 @@ if __name__ == "__main__":
 
     args = get_args()
     compute_fbank_aishell(
-        num_mel_bins=args.num_mel_bins, perturb_speed=args.perturb_speed
+        num_mel_bins=args.num_mel_bins,
+        perturb_speed=args.perturb_speed,
+        whisper_fbank=args.whisper_fbank,
+        output_dir=args.output_dir,
     )

egs/aishell/ASR/prepare.sh

@@ -376,3 +376,16 @@ if [ $stage -le 12 ] && [ $stop_stage -ge 12 ]; then
     --vocab-size 4336 \
     --master-port 12345
 fi
+
+# whisper large-v3 using 128 mel bins, others using 80 mel bins
+whisper_mel_bins=80
+output_dir=data/fbank_whisper
+if [ $stage -le 30 ] && [ $stop_stage -ge 30 ]; then
+  log "Stage 30: Compute ${whisper_mel_bins} dim fbank for whisper model fine-tuning"
+  if [ ! -f $output_dir/.aishell.whisper.done ]; then
+    mkdir -p $output_dir
+    ./local/compute_fbank_aishell.py --perturb-speed ${perturb_speed} --num-mel-bins ${whisper_mel_bins} --whisper-fbank true --output-dir $output_dir
+    ./local/compute_fbank_musan.py --num-mel-bins ${whisper_mel_bins} --whisper-fbank true --output-dir $output_dir
+    touch $output_dir/.aishell.whisper.done
+  fi
+fi

egs/aishell/ASR/pruned_transducer_stateless2/export.py

@@ -47,12 +47,12 @@ import argparse
 import logging
 from pathlib import Path
 
+import k2
 import torch
 from train import add_model_arguments, get_params, get_transducer_model
 
 from icefall.checkpoint import average_checkpoints, find_checkpoints, load_checkpoint
-from icefall.lexicon import Lexicon
+from icefall.utils import num_tokens, str2bool
-from icefall.utils import str2bool
 
 
 def get_parser():
@@ -106,10 +106,10 @@ def get_parser():
     )
 
     parser.add_argument(
-        "--lang-dir",
+        "--tokens",
-        type=Path,
+        type=str,
-        default=Path("data/lang_char"),
+        default="data/lang_char/tokens.txt",
-        help="The lang dir",
+        help="Path to the tokens.txt",
     )
 
     parser.add_argument(
@@ -136,10 +136,9 @@ def main():
 
     logging.info(f"device: {device}")
 
-    lexicon = Lexicon(params.lang_dir)
+    token_table = k2.SymbolTable.from_file(params.tokens)
-
+    params.blank_id = token_table["<blk>"]
-    params.blank_id = 0
+    params.vocab_size = num_tokens(token_table) + 1
-    params.vocab_size = max(lexicon.tokens) + 1
 
     logging.info(params)
 

egs/aishell/ASR/pruned_transducer_stateless2/pretrained.py

@@ -240,6 +240,7 @@ def main():
     opts.frame_opts.snip_edges = False
     opts.frame_opts.samp_freq = params.sample_rate
     opts.mel_opts.num_bins = params.feature_dim
+    opts.mel_opts.high_freq = -400
 
     fbank = kaldifeat.Fbank(opts)
 

egs/aishell/ASR/pruned_transducer_stateless3/export.py

@@ -47,6 +47,7 @@ import argparse
 import logging
 from pathlib import Path
 
+import k2
 import torch
 from scaling_converter import convert_scaled_to_non_scaled
 from train import add_model_arguments, get_params, get_transducer_model
@@ -57,8 +58,7 @@ from icefall.checkpoint import (
     find_checkpoints,
     load_checkpoint,
 )
-from icefall.lexicon import Lexicon
+from icefall.utils import num_tokens, str2bool
-from icefall.utils import str2bool
 
 
 def get_parser():
@@ -123,10 +123,10 @@ def get_parser():
     )
 
     parser.add_argument(
-        "--lang-dir",
+        "--tokens",
-        type=Path,
+        type=str,
-        default=Path("data/lang_char"),
+        default="data/lang_char/tokens.txt",
-        help="The lang dir",
+        help="Path to the tokens.txt",
     )
 
     parser.add_argument(
@@ -153,10 +153,9 @@ def main():
 
     logging.info(f"device: {device}")
 
-    lexicon = Lexicon(params.lang_dir)
+    token_table = k2.SymbolTable.from_file(params.tokens)
-
+    params.blank_id = token_table["<blk>"]
-    params.blank_id = 0
+    params.vocab_size = num_tokens(token_table) + 1
-    params.vocab_size = max(lexicon.tokens) + 1
     params.datatang_prob = 0
 
     logging.info(params)

egs/aishell/ASR/pruned_transducer_stateless3/pretrained.py

@@ -241,6 +241,7 @@ def main():
     opts.frame_opts.snip_edges = False
     opts.frame_opts.samp_freq = params.sample_rate
     opts.mel_opts.num_bins = params.feature_dim
+    opts.mel_opts.high_freq = -400
 
     fbank = kaldifeat.Fbank(opts)
 

egs/aishell/ASR/pruned_transducer_stateless7/export-onnx.py

@@ -49,14 +49,14 @@ import logging
 from pathlib import Path
 from typing import Dict, Tuple
 
+import k2
 import onnx
-import sentencepiece as spm
 import torch
 import torch.nn as nn
 from decoder2 import Decoder
+from do_not_use_it_directly import add_model_arguments, get_params, get_transducer_model
 from onnxruntime.quantization import QuantType, quantize_dynamic
 from scaling_converter import convert_scaled_to_non_scaled
-from do_not_use_it_directly import add_model_arguments, get_params, get_transducer_model
 from zipformer import Zipformer
 
 from icefall.checkpoint import (
@@ -65,8 +65,7 @@ from icefall.checkpoint import (
     find_checkpoints,
     load_checkpoint,
 )
-from icefall.lexicon import Lexicon
+from icefall.utils import num_tokens, setup_logger, str2bool
-from icefall.utils import setup_logger, str2bool
 
 
 def get_parser():
@@ -123,12 +122,10 @@ def get_parser():
     )
 
     parser.add_argument(
-        "--lang-dir",
+        "--tokens",
         type=str,
-        help="""The lang dir
+        default="data/lang_char/tokens.txt",
-        It contains language related input files such as
+        help="Path to the tokens.txt",
-        "lexicon.txt"
-        """,
     )
 
     parser.add_argument(
@@ -404,9 +401,9 @@ def main():
 
     logging.info(f"device: {device}")
 
-    lexicon = Lexicon(params.lang_dir)
+    token_table = k2.SymbolTable.from_file(params.tokens)
-    params.blank_id = 0
+    params.blank_id = token_table["<blk>"]
-    params.vocab_size = max(lexicon.tokens) + 1
+    params.vocab_size = num_tokens(token_table) + 1
 
     logging.info(params)
 

egs/aishell/ASR/pruned_transducer_stateless7/jit_pretrained.py

@@ -230,6 +230,7 @@ def main():
     opts.frame_opts.snip_edges = False
     opts.frame_opts.samp_freq = 16000
     opts.mel_opts.num_bins = 80
+    opts.mel_opts.high_freq = -400
 
     fbank = kaldifeat.Fbank(opts)