From 33d437c5b5baf9d2bde89e409735f7db44f49a50 Mon Sep 17 00:00:00 2001
From: zr_jin <60612200+JinZr@users.noreply.github.com>
Date: Mon, 17 Jul 2023 09:59:54 +0800
Subject: [PATCH] fbank extraction for thchs30

---
 .../ASR/local/compute_fbank_thchs30.py        | 118 ++++++++++++++++++
 egs/multi_zh-hans/ASR/prepare.sh              |  78 ++++++++++++
 2 files changed, 196 insertions(+)
 create mode 100644 egs/multi_zh-hans/ASR/local/compute_fbank_thchs30.py

diff --git a/egs/multi_zh-hans/ASR/local/compute_fbank_thchs30.py b/egs/multi_zh-hans/ASR/local/compute_fbank_thchs30.py
new file mode 100644
index 000000000..202a794af
--- /dev/null
+++ b/egs/multi_zh-hans/ASR/local/compute_fbank_thchs30.py
@@ -0,0 +1,118 @@
+#!/usr/bin/env python3
+# Copyright    2023  Xiaomi Corp.        (authors: Fangjun Kuang
+#                                                  Zengrui Jin)
+#
+# 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.
+
+
+"""
+This file computes fbank features of the aishell dataset.
+It looks for manifests in the directory data/manifests.
+
+The generated fbank features are saved in data/fbank.
+"""
+
+import argparse
+import logging
+import os
+from pathlib import Path
+
+import torch
+from lhotse import CutSet, Fbank, FbankConfig, LilcomChunkyWriter
+from lhotse.recipes.utils import read_manifests_if_cached
+
+from icefall.utils import get_executor
+
+# Torch's multithreaded behavior needs to be disabled or
+# it wastes a lot of CPU and slow things down.
+# Do this outside of main() in case it needs to take effect
+# even when we are not invoking the main (e.g. when spawning subprocesses).
+torch.set_num_threads(1)
+torch.set_num_interop_threads(1)
+
+
+def compute_fbank_thchs30(num_mel_bins: int = 80):
+    src_dir = Path("data/manifests")
+    output_dir = Path("data/fbank")
+    num_jobs = min(15, os.cpu_count())
+
+    dataset_parts = (
+        "train",
+        "dev",
+        "test",
+    )
+    prefix = "thchs30"
+    suffix = "jsonl.gz"
+    manifests = read_manifests_if_cached(
+        dataset_parts=dataset_parts,
+        output_dir=src_dir,
+        prefix=prefix,
+        suffix=suffix,
+    )
+    assert manifests is not None
+
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
+    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():
+            if (output_dir / f"{prefix}_cuts_{partition}.{suffix}").is_file():
+                logging.info(f"{partition} already exists - skipping.")
+                continue
+            logging.info(f"Processing {partition}")
+            cut_set = CutSet.from_manifests(
+                recordings=m["recordings"],
+                supervisions=m["supervisions"],
+            )
+            if "train" in partition:
+                cut_set = (
+                    cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
+                )
+            cut_set = cut_set.compute_and_store_features(
+                extractor=extractor,
+                storage_path=f"{output_dir}/{prefix}_feats_{partition}",
+                # when an executor is specified, make more partitions
+                num_jobs=num_jobs if ex is None else 80,
+                executor=ex,
+                storage_type=LilcomChunkyWriter,
+            )
+            cut_set.to_file(output_dir / f"{prefix}_cuts_{partition}.{suffix}")
+
+
+def get_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--num-mel-bins",
+        type=int,
+        default=80,
+        help="""The number of mel bins for Fbank""",
+    )
+
+    return parser.parse_args()
+
+
+if __name__ == "__main__":
+    formatter = "%(asctime)s %(levelname)s [%(filename)s:%(lineno)d] %(message)s"
+
+    logging.basicConfig(format=formatter, level=logging.INFO)
+
+    args = get_args()
+    compute_fbank_thchs30(num_mel_bins=args.num_mel_bins)
diff --git a/egs/multi_zh-hans/ASR/prepare.sh b/egs/multi_zh-hans/ASR/prepare.sh
index e69de29bb..65b4cc1b8 100755
--- a/egs/multi_zh-hans/ASR/prepare.sh
+++ b/egs/multi_zh-hans/ASR/prepare.sh
@@ -0,0 +1,78 @@
+#!/usr/bin/env bash
+
+# fix segmentation fault reported in https://github.com/k2-fsa/icefall/issues/674
+export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
+
+set -eou pipefail
+
+nj=16
+stage=-1
+stop_stage=100
+
+dl_dir=$PWD/download
+
+. shared/parse_options.sh || exit 1
+
+vocab_sizes=(
+  # 2000
+  # 1000
+  500
+)
+
+
+# multidataset list.
+# LibriSpeech and musan are required.
+# The others are optional.
+multidataset=(
+  "gigaspeech",
+  "commonvoice",
+  "librilight",
+)
+
+# All files generated by this script are saved in "data".
+# You can safely remove "data" and rerun this script to regenerate it.
+mkdir -p data
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+log "dl_dir: $dl_dir"
+
+log "Dataset: musan"
+if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
+  log "Stage 0: Soft link fbank of musan"
+  mkdir -p data/fbank
+  if [ -e ../../librispeech/ASR/data/fbank/.musan.done ]; then
+    cd data/fbank
+    ln -svf $(realpath ../../../../librispeech/ASR/data/fbank/musan_feats) .
+    ln -svf $(realpath ../../../../librispeech/ASR/data/fbank/musan_cuts.jsonl.gz) .
+    cd ../..
+  else
+    log "Abort! Please run ../../librispeech/ASR/prepare.sh --stage 4 --stop-stage 4"
+    exit 1
+  fi
+fi
+
+log "Dataset: THCHS-30"
+if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
+  log "Stage 1: Prepare THCHS-30"
+  if [ ! -d $dl_dir/thchs30 ]; then
+    log "Downloading THCHS-30"
+    lhotse download thchs30 $dl_dir/thchs30
+  fi
+
+  if [ ! -f data/manifests/.thchs30.done ]; then
+    mkdir -p data/manifests
+    lhotse prepare thchs-30 $dl_dir/thchs30 data/manifests/thchs30
+    touch data/manifests/.thchs30.done
+  fi
+
+  if [ ! -f data/fbank/.thchs30.done ]; then
+    mkdir -p data/fbank
+    ./local/compute_fbank_thchs30.py
+    touch data/fbank/.thchs30.done
+  fi
+fi