diff --git a/.github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh b/.github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh index bb7c7dfdc..0bec8c0c4 100755 --- a/.github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh +++ b/.github/scripts/compute-fbank-librispeech-test-clean-and-test-other.sh @@ -15,5 +15,5 @@ mkdir -p data cd data [ ! -e fbank ] && ln -s ~/tmp/fbank-libri fbank cd .. -./local/compute_fbank_librispeech.py +./local/compute_fbank_librispeech.py --dataset 'test-clean test-other' ls -lh data/fbank/ diff --git a/.github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh b/.github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh index e70a1848d..4c393f6be 100755 --- a/.github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh +++ b/.github/scripts/run-aishell-pruned-transducer-stateless3-2022-06-20.sh @@ -25,7 +25,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-conformer-ctc3-2022-11-28.sh b/.github/scripts/run-librispeech-conformer-ctc3-2022-11-28.sh index df29f188e..c68ccc954 100755 --- a/.github/scripts/run-librispeech-conformer-ctc3-2022-11-28.sh +++ b/.github/scripts/run-librispeech-conformer-ctc3-2022-11-28.sh @@ -18,7 +18,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-lstm-transducer-stateless2-2022-09-03.sh b/.github/scripts/run-librispeech-lstm-transducer-stateless2-2022-09-03.sh index 91cdea01a..4cd2c4bec 100755 --- a/.github/scripts/run-librispeech-lstm-transducer-stateless2-2022-09-03.sh +++ b/.github/scripts/run-librispeech-lstm-transducer-stateless2-2022-09-03.sh @@ -20,7 +20,6 @@ abs_repo=$(realpath $repo) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless-2022-03-12.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless-2022-03-12.sh index dafea56db..6792c7088 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless-2022-03-12.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless-2022-03-12.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless2-2022-04-29.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless2-2022-04-29.sh index c3d07dc0e..dbf678d72 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless2-2022-04-29.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless2-2022-04-29.sh @@ -23,7 +23,6 @@ popd log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-04-29.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-04-29.sh index 22de3b45d..b6d477afe 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-04-29.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-04-29.sh @@ -22,7 +22,6 @@ popd log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-05-13.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-05-13.sh index ceb77c7c3..efa4b53f0 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-05-13.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless3-2022-05-13.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless5-2022-05-13.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless5-2022-05-13.sh index c6a781318..511fe0c9e 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless5-2022-05-13.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless5-2022-05-13.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless7-2022-11-11.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless7-2022-11-11.sh index 8e485d2e6..2bc179c86 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless7-2022-11-11.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless7-2022-11-11.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-2022-12-01.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-2022-12-01.sh index 3cbb480f6..192438353 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-2022-12-01.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-2022-12-01.sh @@ -18,7 +18,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp @@ -148,4 +147,4 @@ if [[ x"${GITHUB_EVENT_NAME}" == x"schedule" || x"${GITHUB_EVENT_LABEL_NAME}" == done rm pruned_transducer_stateless7_ctc/exp/*.pt -fi \ No newline at end of file +fi diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-bs-2022-12-15.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-bs-2022-12-15.sh index ed66a728e..761eb72e2 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-bs-2022-12-15.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless7-ctc-bs-2022-12-15.sh @@ -10,7 +10,7 @@ log() { cd egs/librispeech/ASR -repo_url=https://huggingface.co/yfyeung/icefall-asr-librispeech-pruned_transducer_stateless7_ctc_bs-2022-12-14 +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 $repo_url @@ -18,7 +18,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless7-streaming-2022-12-29.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless7-streaming-2022-12-29.sh index 584f5d488..e1e4e1f10 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless7-streaming-2022-12-29.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless7-streaming-2022-12-29.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo diff --git a/.github/scripts/run-librispeech-pruned-transducer-stateless8-2022-11-14.sh b/.github/scripts/run-librispeech-pruned-transducer-stateless8-2022-11-14.sh index e782b8425..5d9485692 100755 --- a/.github/scripts/run-librispeech-pruned-transducer-stateless8-2022-11-14.sh +++ b/.github/scripts/run-librispeech-pruned-transducer-stateless8-2022-11-14.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-streaming-pruned-transducer-stateless2-2022-06-26.sh b/.github/scripts/run-librispeech-streaming-pruned-transducer-stateless2-2022-06-26.sh index af37102d5..77cd59506 100755 --- a/.github/scripts/run-librispeech-streaming-pruned-transducer-stateless2-2022-06-26.sh +++ b/.github/scripts/run-librispeech-streaming-pruned-transducer-stateless2-2022-06-26.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-librispeech-transducer-stateless2-2022-04-19.sh b/.github/scripts/run-librispeech-transducer-stateless2-2022-04-19.sh index 5b8ed396b..b4aca1b6b 100755 --- a/.github/scripts/run-librispeech-transducer-stateless2-2022-04-19.sh +++ b/.github/scripts/run-librispeech-transducer-stateless2-2022-04-19.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-librispeech-zipformer-mmi-2022-12-08.sh b/.github/scripts/run-librispeech-zipformer-mmi-2022-12-08.sh index 77f28b054..a58b8ec56 100755 --- a/.github/scripts/run-librispeech-zipformer-mmi-2022-12-08.sh +++ b/.github/scripts/run-librispeech-zipformer-mmi-2022-12-08.sh @@ -18,7 +18,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/run-pre-trained-conformer-ctc.sh b/.github/scripts/run-pre-trained-conformer-ctc.sh index 96c320616..125d1f3b1 100755 --- a/.github/scripts/run-pre-trained-conformer-ctc.sh +++ b/.github/scripts/run-pre-trained-conformer-ctc.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.flac ls -lh $repo/test_wavs/*.flac log "CTC decoding" diff --git a/.github/scripts/run-pre-trained-transducer-stateless-librispeech-100h.sh b/.github/scripts/run-pre-trained-transducer-stateless-librispeech-100h.sh index 209d4814f..89115e88d 100755 --- a/.github/scripts/run-pre-trained-transducer-stateless-librispeech-100h.sh +++ b/.github/scripts/run-pre-trained-transducer-stateless-librispeech-100h.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-pre-trained-transducer-stateless-librispeech-960h.sh b/.github/scripts/run-pre-trained-transducer-stateless-librispeech-960h.sh index 34ff76fe4..85e2c89e6 100755 --- a/.github/scripts/run-pre-trained-transducer-stateless-librispeech-960h.sh +++ b/.github/scripts/run-pre-trained-transducer-stateless-librispeech-960h.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-pre-trained-transducer-stateless-modified-2-aishell.sh b/.github/scripts/run-pre-trained-transducer-stateless-modified-2-aishell.sh index 75650c2d3..0644d9be0 100755 --- a/.github/scripts/run-pre-trained-transducer-stateless-modified-2-aishell.sh +++ b/.github/scripts/run-pre-trained-transducer-stateless-modified-2-aishell.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-pre-trained-transducer-stateless-modified-aishell.sh b/.github/scripts/run-pre-trained-transducer-stateless-modified-aishell.sh index bcc2d74cb..79fb64311 100755 --- a/.github/scripts/run-pre-trained-transducer-stateless-modified-aishell.sh +++ b/.github/scripts/run-pre-trained-transducer-stateless-modified-aishell.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-pre-trained-transducer-stateless.sh b/.github/scripts/run-pre-trained-transducer-stateless.sh index d3e40315a..41456f11b 100755 --- a/.github/scripts/run-pre-trained-transducer-stateless.sh +++ b/.github/scripts/run-pre-trained-transducer-stateless.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav for sym in 1 2 3; do diff --git a/.github/scripts/run-pre-trained-transducer.sh b/.github/scripts/run-pre-trained-transducer.sh index cfa006776..1331c966c 100755 --- a/.github/scripts/run-pre-trained-transducer.sh +++ b/.github/scripts/run-pre-trained-transducer.sh @@ -19,7 +19,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav log "Beam search decoding" diff --git a/.github/scripts/run-wenetspeech-pruned-transducer-stateless2.sh b/.github/scripts/run-wenetspeech-pruned-transducer-stateless2.sh index 2d237dcf2..90097c752 100755 --- a/.github/scripts/run-wenetspeech-pruned-transducer-stateless2.sh +++ b/.github/scripts/run-wenetspeech-pruned-transducer-stateless2.sh @@ -20,7 +20,6 @@ repo=$(basename $repo_url) log "Display test files" tree $repo/ -soxi $repo/test_wavs/*.wav ls -lh $repo/test_wavs/*.wav pushd $repo/exp diff --git a/.github/scripts/test-ncnn-export.sh b/.github/scripts/test-ncnn-export.sh index 9f5df2d58..52491d2ea 100755 --- a/.github/scripts/test-ncnn-export.sh +++ b/.github/scripts/test-ncnn-export.sh @@ -232,70 +232,3 @@ python3 ./pruned_transducer_stateless7_streaming/streaming-ncnn-decode.py \ rm -rf $repo log "--------------------------------------------------------------------------" - -# Go back to the root directory of icefall repo -popd - -pushd egs/csj/ASR - -log "==========================================================================" -repo_url=https://huggingface.co/TeoWenShen/icefall-asr-csj-pruned-transducer-stateless7-streaming-230208 -GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url -repo=$(basename $repo_url) - -pushd $repo -git lfs pull --include "exp_fluent/pretrained.pt" -git lfs pull --include "exp_disfluent/pretrained.pt" - -cd exp_fluent -ln -s pretrained.pt epoch-99.pt - -cd ../exp_disfluent -ln -s pretrained.pt epoch-99.pt - -cd ../test_wavs -git lfs pull --include "*.wav" -popd - -log "Export via torch.jit.trace()" - -for exp in exp_fluent exp_disfluent; do - ./pruned_transducer_stateless7_streaming/export-for-ncnn.py \ - --exp-dir $repo/$exp/ \ - --lang $repo/data/lang_char \ - --epoch 99 \ - --avg 1 \ - --use-averaged-model 0 \ - \ - --decode-chunk-len 32 \ - --num-left-chunks 4 \ - --num-encoder-layers "2,4,3,2,4" \ - --feedforward-dims "1024,1024,2048,2048,1024" \ - --nhead "8,8,8,8,8" \ - --encoder-dims "384,384,384,384,384" \ - --attention-dims "192,192,192,192,192" \ - --encoder-unmasked-dims "256,256,256,256,256" \ - --zipformer-downsampling-factors "1,2,4,8,2" \ - --cnn-module-kernels "31,31,31,31,31" \ - --decoder-dim 512 \ - --joiner-dim 512 - - pnnx $repo/$exp/encoder_jit_trace-pnnx.pt - pnnx $repo/$exp/decoder_jit_trace-pnnx.pt - pnnx $repo/$exp/joiner_jit_trace-pnnx.pt - - for wav in aps-smp.wav interview_aps-smp.wav reproduction-smp.wav sps-smp.wav; do - python3 ./pruned_transducer_stateless7_streaming/streaming-ncnn-decode.py \ - --tokens $repo/data/lang_char/tokens.txt \ - --encoder-param-filename $repo/$exp/encoder_jit_trace-pnnx.ncnn.param \ - --encoder-bin-filename $repo/$exp/encoder_jit_trace-pnnx.ncnn.bin \ - --decoder-param-filename $repo/$exp/decoder_jit_trace-pnnx.ncnn.param \ - --decoder-bin-filename $repo/$exp/decoder_jit_trace-pnnx.ncnn.bin \ - --joiner-param-filename $repo/$exp/joiner_jit_trace-pnnx.ncnn.param \ - --joiner-bin-filename $repo/$exp/joiner_jit_trace-pnnx.ncnn.bin \ - $repo/test_wavs/$wav - done -done - -rm -rf $repo -log "--------------------------------------------------------------------------" diff --git a/.github/workflows/run-aishell-2022-06-20.yml b/.github/workflows/run-aishell-2022-06-20.yml index 1865a0da8..f5ba73195 100644 --- a/.github/workflows/run-aishell-2022-06-20.yml +++ b/.github/workflows/run-aishell-2022-06-20.yml @@ -65,7 +65,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -87,7 +87,7 @@ jobs: GITHUB_EVENT_NAME: ${{ github.event_name }} GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }} run: | - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-gigaspeech-2022-05-13.yml b/.github/workflows/run-gigaspeech-2022-05-13.yml index e438c5dba..c7b9cc79d 100644 --- a/.github/workflows/run-gigaspeech-2022-05-13.yml +++ b/.github/workflows/run-gigaspeech-2022-05-13.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache diff --git a/.github/workflows/run-librispeech-2022-03-12.yml b/.github/workflows/run-librispeech-2022-03-12.yml index 3ba6850cd..9c7cd1228 100644 --- a/.github/workflows/run-librispeech-2022-03-12.yml +++ b/.github/workflows/run-librispeech-2022-03-12.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-04-29.yml b/.github/workflows/run-librispeech-2022-04-29.yml index 595b410b8..78c9e759f 100644 --- a/.github/workflows/run-librispeech-2022-04-29.yml +++ b/.github/workflows/run-librispeech-2022-04-29.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-05-13.yml b/.github/workflows/run-librispeech-2022-05-13.yml index eb0b06a2d..04799bf52 100644 --- a/.github/workflows/run-librispeech-2022-05-13.yml +++ b/.github/workflows/run-librispeech-2022-05-13.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-11-11-stateless7.yml b/.github/workflows/run-librispeech-2022-11-11-stateless7.yml index 365e2761a..6dfc23920 100644 --- a/.github/workflows/run-librispeech-2022-11-11-stateless7.yml +++ b/.github/workflows/run-librispeech-2022-11-11-stateless7.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-11-14-stateless8.yml b/.github/workflows/run-librispeech-2022-11-14-stateless8.yml index acb11a8f4..0544e68b3 100644 --- a/.github/workflows/run-librispeech-2022-11-14-stateless8.yml +++ b/.github/workflows/run-librispeech-2022-11-14-stateless8.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-12-01-stateless7-ctc.yml b/.github/workflows/run-librispeech-2022-12-01-stateless7-ctc.yml index ccd8d50d0..62e1f2a01 100644 --- a/.github/workflows/run-librispeech-2022-12-01-stateless7-ctc.yml +++ b/.github/workflows/run-librispeech-2022-12-01-stateless7-ctc.yml @@ -60,7 +60,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -119,7 +119,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-12-08-zipformer-mmi.yml b/.github/workflows/run-librispeech-2022-12-08-zipformer-mmi.yml index 5472ca59b..7dc33aaa9 100644 --- a/.github/workflows/run-librispeech-2022-12-08-zipformer-mmi.yml +++ b/.github/workflows/run-librispeech-2022-12-08-zipformer-mmi.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-12-15-stateless7-ctc-bs.yml b/.github/workflows/run-librispeech-2022-12-15-stateless7-ctc-bs.yml index 6e2b40cf3..de55847ad 100644 --- a/.github/workflows/run-librispeech-2022-12-15-stateless7-ctc-bs.yml +++ b/.github/workflows/run-librispeech-2022-12-15-stateless7-ctc-bs.yml @@ -35,7 +35,7 @@ on: jobs: run_librispeech_2022_12_15_zipformer_ctc_bs: - if: github.event.label.name == 'ready' || github.event.label.name == 'run-decode' || github.event.label.name == 'blank-skip' || github.event_name == 'push' || github.event_name == 'schedule' + if: github.event.label.name == 'run-decode' || github.event.label.name == 'blank-skip' || github.event_name == 'push' || github.event_name == 'schedule' runs-on: ${{ matrix.os }} strategy: matrix: @@ -60,7 +60,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -119,7 +119,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-2022-12-29-stateless7-streaming.yml b/.github/workflows/run-librispeech-2022-12-29-stateless7-streaming.yml index 6dd93946a..feb5c6fd0 100644 --- a/.github/workflows/run-librispeech-2022-12-29-stateless7-streaming.yml +++ b/.github/workflows/run-librispeech-2022-12-29-stateless7-streaming.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-conformer-ctc3-2022-11-28.yml b/.github/workflows/run-librispeech-conformer-ctc3-2022-11-28.yml index d763fb1c5..c95ed8b9a 100644 --- a/.github/workflows/run-librispeech-conformer-ctc3-2022-11-28.yml +++ b/.github/workflows/run-librispeech-conformer-ctc3-2022-11-28.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-lstm-transducer-stateless2-2022-09-03.yml b/.github/workflows/run-librispeech-lstm-transducer-stateless2-2022-09-03.yml index f737d9a25..e14d4e92f 100644 --- a/.github/workflows/run-librispeech-lstm-transducer-stateless2-2022-09-03.yml +++ b/.github/workflows/run-librispeech-lstm-transducer-stateless2-2022-09-03.yml @@ -47,7 +47,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -106,7 +106,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-pruned-transducer-stateless3-2022-05-13.yml b/.github/workflows/run-librispeech-pruned-transducer-stateless3-2022-05-13.yml index f67f7599b..73d91fcd4 100644 --- a/.github/workflows/run-librispeech-pruned-transducer-stateless3-2022-05-13.yml +++ b/.github/workflows/run-librispeech-pruned-transducer-stateless3-2022-05-13.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-streaming-transducer-stateless2-2022-06-26.yml b/.github/workflows/run-librispeech-streaming-transducer-stateless2-2022-06-26.yml index ac7e58b20..8a690393e 100644 --- a/.github/workflows/run-librispeech-streaming-transducer-stateless2-2022-06-26.yml +++ b/.github/workflows/run-librispeech-streaming-transducer-stateless2-2022-06-26.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-librispeech-transducer-stateless2-2022-04-19.yml b/.github/workflows/run-librispeech-transducer-stateless2-2022-04-19.yml index 575727e22..217dbdfa1 100644 --- a/.github/workflows/run-librispeech-transducer-stateless2-2022-04-19.yml +++ b/.github/workflows/run-librispeech-transducer-stateless2-2022-04-19.yml @@ -64,7 +64,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -123,7 +123,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-conformer-ctc.yml b/.github/workflows/run-pretrained-conformer-ctc.yml index 7dbfd2bd9..4e8e7b8db 100644 --- a/.github/workflows/run-pretrained-conformer-ctc.yml +++ b/.github/workflows/run-pretrained-conformer-ctc.yml @@ -54,7 +54,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -73,7 +73,7 @@ jobs: - name: Inference with pre-trained model shell: bash run: | - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-transducer-stateless-librispeech-100h.yml b/.github/workflows/run-pretrained-transducer-stateless-librispeech-100h.yml index d6b3de8d4..ddde4f1d6 100644 --- a/.github/workflows/run-pretrained-transducer-stateless-librispeech-100h.yml +++ b/.github/workflows/run-pretrained-transducer-stateless-librispeech-100h.yml @@ -63,7 +63,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -122,7 +122,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-transducer-stateless-librispeech-multi-datasets.yml b/.github/workflows/run-pretrained-transducer-stateless-librispeech-multi-datasets.yml index 749fb3fca..00ea97b2a 100644 --- a/.github/workflows/run-pretrained-transducer-stateless-librispeech-multi-datasets.yml +++ b/.github/workflows/run-pretrained-transducer-stateless-librispeech-multi-datasets.yml @@ -63,7 +63,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -122,7 +122,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-transducer-stateless-modified-2-aishell.yml b/.github/workflows/run-pretrained-transducer-stateless-modified-2-aishell.yml index 92bf6feb8..b3cfc9efd 100644 --- a/.github/workflows/run-pretrained-transducer-stateless-modified-2-aishell.yml +++ b/.github/workflows/run-pretrained-transducer-stateless-modified-2-aishell.yml @@ -54,7 +54,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -73,7 +73,7 @@ jobs: - name: Inference with pre-trained model shell: bash run: | - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-transducer-stateless-modified-aishell.yml b/.github/workflows/run-pretrained-transducer-stateless-modified-aishell.yml index e51da8bd8..ab598541d 100644 --- a/.github/workflows/run-pretrained-transducer-stateless-modified-aishell.yml +++ b/.github/workflows/run-pretrained-transducer-stateless-modified-aishell.yml @@ -54,7 +54,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -73,7 +73,7 @@ jobs: - name: Inference with pre-trained model shell: bash run: | - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-transducer-stateless.yml b/.github/workflows/run-pretrained-transducer-stateless.yml index 2103d0510..d663d49dd 100644 --- a/.github/workflows/run-pretrained-transducer-stateless.yml +++ b/.github/workflows/run-pretrained-transducer-stateless.yml @@ -63,7 +63,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -122,7 +122,7 @@ jobs: ln -sfv ~/tmp/fbank-libri egs/librispeech/ASR/data/fbank ls -lh egs/librispeech/ASR/data/* - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-pretrained-transducer.yml b/.github/workflows/run-pretrained-transducer.yml index 902319b55..9cb9d3b59 100644 --- a/.github/workflows/run-pretrained-transducer.yml +++ b/.github/workflows/run-pretrained-transducer.yml @@ -54,7 +54,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -73,7 +73,7 @@ jobs: - name: Inference with pre-trained model shell: bash run: | - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/run-ptb-rnn-lm.yml b/.github/workflows/run-ptb-rnn-lm.yml index 47ed958f2..f8d9c02c5 100644 --- a/.github/workflows/run-ptb-rnn-lm.yml +++ b/.github/workflows/run-ptb-rnn-lm.yml @@ -47,7 +47,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | grep -v kaldifst | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Prepare data shell: bash diff --git a/.github/workflows/run-wenetspeech-pruned-transducer-stateless2.yml b/.github/workflows/run-wenetspeech-pruned-transducer-stateless2.yml index 8a7be0b80..14fb96ec8 100644 --- a/.github/workflows/run-wenetspeech-pruned-transducer-stateless2.yml +++ b/.github/workflows/run-wenetspeech-pruned-transducer-stateless2.yml @@ -54,7 +54,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache @@ -76,7 +76,7 @@ jobs: GITHUB_EVENT_NAME: ${{ github.event_name }} GITHUB_EVENT_LABEL_NAME: ${{ github.event.label.name }} run: | - sudo apt-get -qq install git-lfs tree sox + sudo apt-get -qq install git-lfs tree export PYTHONPATH=$PWD:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/kaldifeat/python:$PYTHONPATH export PYTHONPATH=~/tmp/kaldifeat/build/lib:$PYTHONPATH diff --git a/.github/workflows/test-ncnn-export.yml b/.github/workflows/test-ncnn-export.yml index e10cfe76b..cdea54854 100644 --- a/.github/workflows/test-ncnn-export.yml +++ b/.github/workflows/test-ncnn-export.yml @@ -46,7 +46,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache diff --git a/.github/workflows/test-onnx-export.yml b/.github/workflows/test-onnx-export.yml index c7729dedb..3dc4261ab 100644 --- a/.github/workflows/test-onnx-export.yml +++ b/.github/workflows/test-onnx-export.yml @@ -46,7 +46,7 @@ jobs: run: | grep -v '^#' ./requirements-ci.txt | xargs -n 1 -L 1 pip install pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* - name: Cache kaldifeat id: my-cache diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml index c062a2a3d..0da4f6b4b 100644 --- a/.github/workflows/test.yml +++ b/.github/workflows/test.yml @@ -56,7 +56,7 @@ jobs: 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 + sudo apt install -q -y --fix-missing libsox-dev libsox-fmt-all - name: Install Python dependencies run: | @@ -70,7 +70,7 @@ jobs: pip install git+https://github.com/lhotse-speech/lhotse # icefall requirements pip uninstall -y protobuf - pip install --no-binary protobuf protobuf + pip install --no-binary protobuf protobuf==3.20.* pip install kaldifst pip install onnxruntime diff --git a/egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/decode.py b/egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/decode.py index d0f118959..2512f233f 100755 --- a/egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/decode.py @@ -391,18 +391,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -412,9 +408,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell/ASR/pruned_transducer_stateless2/decode.py b/egs/aishell/ASR/pruned_transducer_stateless2/decode.py index 20a4f21c7..fb6c7c481 100755 --- a/egs/aishell/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/aishell/ASR/pruned_transducer_stateless2/decode.py @@ -388,18 +388,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" # we compute CER for aishell dataset. results_char = [] for res in results: @@ -413,9 +409,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell/ASR/pruned_transducer_stateless3/decode.py b/egs/aishell/ASR/pruned_transducer_stateless3/decode.py index bac829ae1..954d9dc7e 100755 --- a/egs/aishell/ASR/pruned_transducer_stateless3/decode.py +++ b/egs/aishell/ASR/pruned_transducer_stateless3/decode.py @@ -406,18 +406,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" # we compute CER for aishell dataset. results_char = [] for res in results: @@ -431,9 +427,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tCER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell/ASR/transducer_stateless/decode.py b/egs/aishell/ASR/transducer_stateless/decode.py index e019d2329..d23f4f883 100755 --- a/egs/aishell/ASR/transducer_stateless/decode.py +++ b/egs/aishell/ASR/transducer_stateless/decode.py @@ -325,17 +325,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" # we compute CER for aishell dataset. results_char = [] for res in results: @@ -349,9 +345,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tCER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell/ASR/transducer_stateless_modified-2/decode.py b/egs/aishell/ASR/transducer_stateless_modified-2/decode.py index 41cc1c01c..d164b6890 100755 --- a/egs/aishell/ASR/transducer_stateless_modified-2/decode.py +++ b/egs/aishell/ASR/transducer_stateless_modified-2/decode.py @@ -370,18 +370,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" # we compute CER for aishell dataset. results_char = [] for res in results: @@ -395,9 +391,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tCER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell/ASR/transducer_stateless_modified/decode.py b/egs/aishell/ASR/transducer_stateless_modified/decode.py index 7c06e6e51..0a7d87fe8 100755 --- a/egs/aishell/ASR/transducer_stateless_modified/decode.py +++ b/egs/aishell/ASR/transducer_stateless_modified/decode.py @@ -374,18 +374,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" # we compute CER for aishell dataset. results_char = [] for res in results: @@ -399,9 +395,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tCER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell2/ASR/pruned_transducer_stateless5/decode.py b/egs/aishell2/ASR/pruned_transducer_stateless5/decode.py index b5da0959b..9e44b4e34 100755 --- a/egs/aishell2/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/aishell2/ASR/pruned_transducer_stateless5/decode.py @@ -543,18 +543,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -564,9 +560,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/aishell4/ASR/pruned_transducer_stateless5/decode.py b/egs/aishell4/ASR/pruned_transducer_stateless5/decode.py index 37d766ec8..068e2749a 100755 --- a/egs/aishell4/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/aishell4/ASR/pruned_transducer_stateless5/decode.py @@ -406,18 +406,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -427,9 +423,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/alimeeting/ASR/pruned_transducer_stateless2/decode.py b/egs/alimeeting/ASR/pruned_transducer_stateless2/decode.py index e4a90ef71..6c170c392 100755 --- a/egs/alimeeting/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/alimeeting/ASR/pruned_transducer_stateless2/decode.py @@ -391,18 +391,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -412,9 +408,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/alimeeting/ASR_v2/pruned_transducer_stateless7/decode.py b/egs/alimeeting/ASR_v2/pruned_transducer_stateless7/decode.py index 53381c1f4..2741e0eeb 100755 --- a/egs/alimeeting/ASR_v2/pruned_transducer_stateless7/decode.py +++ b/egs/alimeeting/ASR_v2/pruned_transducer_stateless7/decode.py @@ -462,18 +462,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -483,9 +479,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/ami/ASR/pruned_transducer_stateless7/decode.py b/egs/ami/ASR/pruned_transducer_stateless7/decode.py index f47228fbe..9999894d1 100755 --- a/egs/ami/ASR/pruned_transducer_stateless7/decode.py +++ b/egs/ami/ASR/pruned_transducer_stateless7/decode.py @@ -478,17 +478,13 @@ def save_results( test_set_wers = dict() test_set_cers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - wers_filename = ( - params.res_dir / f"wers-{test_set_name}-{key}-{params.suffix}.txt" - ) + wers_filename = params.res_dir / f"wers-{test_set_name}-{params.suffix}.txt" with open(wers_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -499,9 +495,7 @@ def save_results( results_char = [] for res in results: results_char.append((res[0], list("".join(res[1])), list("".join(res[2])))) - cers_filename = ( - params.res_dir / f"cers-{test_set_name}-{key}-{params.suffix}.txt" - ) + cers_filename = params.res_dir / f"cers-{test_set_name}-{params.suffix}.txt" with open(cers_filename, "w") as f: cer = write_error_stats( f, f"{test_set_name}-{key}", results_char, enable_log=True @@ -512,9 +506,7 @@ def save_results( test_set_wers = {k: v for k, v in sorted(test_set_wers.items(), key=lambda x: x[1])} test_set_cers = {k: v for k, v in sorted(test_set_cers.items(), key=lambda x: x[1])} - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER\tCER", file=f) for key in test_set_wers: diff --git a/egs/csj/ASR/pruned_transducer_stateless7_streaming/decode.py b/egs/csj/ASR/pruned_transducer_stateless7_streaming/decode.py index 19d3c79c8..f5a1d750d 100755 --- a/egs/csj/ASR/pruned_transducer_stateless7_streaming/decode.py +++ b/egs/csj/ASR/pruned_transducer_stateless7_streaming/decode.py @@ -599,9 +599,7 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) @@ -609,9 +607,7 @@ def save_results( # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -621,9 +617,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/gigaspeech/ASR/pruned_transducer_stateless2/decode.py b/egs/gigaspeech/ASR/pruned_transducer_stateless2/decode.py index 8595c27bd..ee694a9e0 100755 --- a/egs/gigaspeech/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/gigaspeech/ASR/pruned_transducer_stateless2/decode.py @@ -399,9 +399,7 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = post_processing(results) results = sorted(results) store_transcripts(filename=recog_path, texts=results) @@ -409,9 +407,7 @@ def save_results( # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -421,9 +417,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/conformer_ctc3/decode.py b/egs/librispeech/ASR/conformer_ctc3/decode.py index 6fbf9d674..e6327bb5e 100755 --- a/egs/librispeech/ASR/conformer_ctc3/decode.py +++ b/egs/librispeech/ASR/conformer_ctc3/decode.py @@ -728,18 +728,14 @@ def save_results( test_set_wers = dict() test_set_delays = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts_and_timestamps(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer, mean_delay, var_delay = write_error_stats_with_timestamps( f, @@ -754,9 +750,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: @@ -765,8 +759,7 @@ def save_results( # sort according to the mean start symbol delay test_set_delays = sorted(test_set_delays.items(), key=lambda x: x[1][0][0]) delays_info = ( - params.res_dir - / f"symbol-delay-summary-{test_set_name}-{key}-{params.suffix}.txt" + params.res_dir / f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt" ) with open(delays_info, "w") as f: print("settings\t(start, end) symbol-delay (s) (start, end)", file=f) diff --git a/egs/librispeech/ASR/conv_emformer_transducer_stateless/decode.py b/egs/librispeech/ASR/conv_emformer_transducer_stateless/decode.py index 365e8b8a7..7be3299f3 100755 --- a/egs/librispeech/ASR/conv_emformer_transducer_stateless/decode.py +++ b/egs/librispeech/ASR/conv_emformer_transducer_stateless/decode.py @@ -432,18 +432,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -453,9 +449,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/conv_emformer_transducer_stateless/streaming_decode.py b/egs/librispeech/ASR/conv_emformer_transducer_stateless/streaming_decode.py index c93125c80..e5a7c7116 100755 --- a/egs/librispeech/ASR/conv_emformer_transducer_stateless/streaming_decode.py +++ b/egs/librispeech/ASR/conv_emformer_transducer_stateless/streaming_decode.py @@ -750,17 +750,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=sorted(results)) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -770,9 +766,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/conv_emformer_transducer_stateless2/decode.py b/egs/librispeech/ASR/conv_emformer_transducer_stateless2/decode.py index 78e1f4096..d022d463e 100755 --- a/egs/librispeech/ASR/conv_emformer_transducer_stateless2/decode.py +++ b/egs/librispeech/ASR/conv_emformer_transducer_stateless2/decode.py @@ -432,18 +432,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -453,9 +449,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/conv_emformer_transducer_stateless2/streaming_decode.py b/egs/librispeech/ASR/conv_emformer_transducer_stateless2/streaming_decode.py index b2cb2c96b..f5d894a7b 100755 --- a/egs/librispeech/ASR/conv_emformer_transducer_stateless2/streaming_decode.py +++ b/egs/librispeech/ASR/conv_emformer_transducer_stateless2/streaming_decode.py @@ -750,17 +750,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=sorted(results)) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -770,9 +766,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/local/compute_fbank_librispeech.py b/egs/librispeech/ASR/local/compute_fbank_librispeech.py index 9f8503814..745eaf1e8 100755 --- a/egs/librispeech/ASR/local/compute_fbank_librispeech.py +++ b/egs/librispeech/ASR/local/compute_fbank_librispeech.py @@ -54,10 +54,20 @@ def get_args(): help="""Path to the bpe.model. If not None, we will remove short and long utterances before extracting features""", ) + + parser.add_argument( + "--dataset", + type=str, + help="""Dataset parts to compute fbank. If None, we will use all""", + ) + return parser.parse_args() -def compute_fbank_librispeech(bpe_model: Optional[str] = None): +def compute_fbank_librispeech( + bpe_model: Optional[str] = None, + dataset: Optional[str] = None, +): src_dir = Path("data/manifests") output_dir = Path("data/fbank") num_jobs = min(15, os.cpu_count()) @@ -68,15 +78,19 @@ def compute_fbank_librispeech(bpe_model: Optional[str] = None): sp = spm.SentencePieceProcessor() sp.load(bpe_model) - dataset_parts = ( - "dev-clean", - "dev-other", - "test-clean", - "test-other", - "train-clean-100", - "train-clean-360", - "train-other-500", - ) + if dataset is None: + dataset_parts = ( + "dev-clean", + "dev-other", + "test-clean", + "test-other", + "train-clean-100", + "train-clean-360", + "train-other-500", + ) + else: + dataset_parts = dataset.split(" ", -1) + prefix = "librispeech" suffix = "jsonl.gz" manifests = read_manifests_if_cached( @@ -131,4 +145,4 @@ if __name__ == "__main__": logging.basicConfig(format=formatter, level=logging.INFO) args = get_args() logging.info(vars(args)) - compute_fbank_librispeech(bpe_model=args.bpe_model) + compute_fbank_librispeech(bpe_model=args.bpe_model, dataset=args.dataset) diff --git a/egs/librispeech/ASR/lstm_transducer_stateless/decode.py b/egs/librispeech/ASR/lstm_transducer_stateless/decode.py index 3ad08f56a..856c9d945 100755 --- a/egs/librispeech/ASR/lstm_transducer_stateless/decode.py +++ b/egs/librispeech/ASR/lstm_transducer_stateless/decode.py @@ -566,18 +566,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -587,9 +583,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/lstm_transducer_stateless/streaming_decode.py b/egs/librispeech/ASR/lstm_transducer_stateless/streaming_decode.py index 961d8ddfb..f989d9bc0 100755 --- a/egs/librispeech/ASR/lstm_transducer_stateless/streaming_decode.py +++ b/egs/librispeech/ASR/lstm_transducer_stateless/streaming_decode.py @@ -742,17 +742,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=sorted(results)) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -762,9 +758,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/lstm_transducer_stateless2/decode.py b/egs/librispeech/ASR/lstm_transducer_stateless2/decode.py index 78be9c01f..6c58a57e1 100755 --- a/egs/librispeech/ASR/lstm_transducer_stateless2/decode.py +++ b/egs/librispeech/ASR/lstm_transducer_stateless2/decode.py @@ -702,18 +702,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -723,9 +719,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/lstm_transducer_stateless3/decode.py b/egs/librispeech/ASR/lstm_transducer_stateless3/decode.py index a380bc470..a2b4f9e1a 100755 --- a/egs/librispeech/ASR/lstm_transducer_stateless3/decode.py +++ b/egs/librispeech/ASR/lstm_transducer_stateless3/decode.py @@ -611,18 +611,14 @@ def save_results( test_set_wers = dict() test_set_delays = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts_and_timestamps(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer, mean_delay, var_delay = write_error_stats_with_timestamps( f, f"{test_set_name}-{key}", results, enable_log=True @@ -633,9 +629,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: @@ -643,8 +637,7 @@ def save_results( test_set_delays = sorted(test_set_delays.items(), key=lambda x: x[1][0]) delays_info = ( - params.res_dir - / f"symbol-delay-summary-{test_set_name}-{key}-{params.suffix}.txt" + params.res_dir / f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt" ) with open(delays_info, "w") as f: print("settings\tsymbol-delay", file=f) diff --git a/egs/librispeech/ASR/lstm_transducer_stateless3/streaming_decode.py b/egs/librispeech/ASR/lstm_transducer_stateless3/streaming_decode.py index 109746ed5..c737e3611 100755 --- a/egs/librispeech/ASR/lstm_transducer_stateless3/streaming_decode.py +++ b/egs/librispeech/ASR/lstm_transducer_stateless3/streaming_decode.py @@ -742,17 +742,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=sorted(results)) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -762,9 +758,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned2_knowledge/decode.py b/egs/librispeech/ASR/pruned2_knowledge/decode.py index 40d14bb5a..82fd103ea 100755 --- a/egs/librispeech/ASR/pruned2_knowledge/decode.py +++ b/egs/librispeech/ASR/pruned2_knowledge/decode.py @@ -386,17 +386,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -406,9 +402,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_stateless_emformer_rnnt2/decode.py b/egs/librispeech/ASR/pruned_stateless_emformer_rnnt2/decode.py index 0e3b7ff74..072d49d9c 100755 --- a/egs/librispeech/ASR/pruned_stateless_emformer_rnnt2/decode.py +++ b/egs/librispeech/ASR/pruned_stateless_emformer_rnnt2/decode.py @@ -420,18 +420,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -441,9 +437,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless/decode.py index 0444afe40..6dfe11cee 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless/decode.py @@ -585,18 +585,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -606,9 +602,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless/streaming_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless/streaming_decode.py index fbc39fb65..f4b01fd06 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless/streaming_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless/streaming_decode.py @@ -423,9 +423,7 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" # sort results so we can easily compare the difference between two # recognition results results = sorted(results) @@ -434,9 +432,7 @@ def save_results( # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -446,9 +442,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless2/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless2/decode.py index 5f135f219..172c9ab7c 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless2/decode.py @@ -609,18 +609,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -630,9 +626,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless2/streaming_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless2/streaming_decode.py index bb08246d9..9c4a13606 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless2/streaming_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless2/streaming_decode.py @@ -425,9 +425,7 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" # sort results so we can easily compare the difference between two # recognition results results = sorted(results) @@ -436,9 +434,7 @@ def save_results( # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -448,9 +444,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless3/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless3/decode.py index 109a94a69..aa055049e 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless3/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless3/decode.py @@ -869,18 +869,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -890,9 +886,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless3/streaming_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless3/streaming_decode.py index 0e5111f33..3a1ecb7ed 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless3/streaming_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless3/streaming_decode.py @@ -426,18 +426,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -447,9 +443,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless4/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless4/decode.py index c44db0206..5ec3d3b45 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless4/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless4/decode.py @@ -655,18 +655,14 @@ def save_results( test_set_wers = dict() test_set_delays = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts_and_timestamps(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer, mean_delay, var_delay = write_error_stats_with_timestamps( f, f"{test_set_name}-{key}", results, enable_log=True @@ -677,9 +673,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: @@ -687,8 +681,7 @@ def save_results( test_set_delays = sorted(test_set_delays.items(), key=lambda x: x[1][0]) delays_info = ( - params.res_dir - / f"symbol-delay-summary-{test_set_name}-{key}-{params.suffix}.txt" + params.res_dir / f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt" ) with open(delays_info, "w") as f: print("settings\tsymbol-delay", file=f) diff --git a/egs/librispeech/ASR/pruned_transducer_stateless4/streaming_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless4/streaming_decode.py index c4e3cef16..ca3a023ce 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless4/streaming_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless4/streaming_decode.py @@ -442,18 +442,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -463,9 +459,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless5/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless5/decode.py index 90b0fcf4b..2be895feb 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless5/decode.py @@ -735,18 +735,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -756,9 +752,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless5/streaming_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless5/streaming_decode.py index 064811f1c..5b15dcee7 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless5/streaming_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless5/streaming_decode.py @@ -442,18 +442,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -463,9 +459,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless6/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless6/decode.py index fd9de052a..95534efef 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless6/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless6/decode.py @@ -416,18 +416,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -437,9 +433,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7/decode.py index b9bce465f..32b3134b9 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7/decode.py @@ -722,18 +722,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -743,9 +739,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/ctc_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/ctc_decode.py index 4b373e4c7..629bec058 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/ctc_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/ctc_decode.py @@ -541,18 +541,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats(f, f"{test_set_name}-{key}", results) test_set_wers[key] = wer @@ -560,9 +556,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/decode.py index 32a9b6bb2..7641fa5af 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc/decode.py @@ -593,18 +593,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -614,9 +610,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/ctc_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/ctc_decode.py index f137485b2..fa7144f0f 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/ctc_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/ctc_decode.py @@ -532,18 +532,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats(f, f"{test_set_name}-{key}", results) test_set_wers[key] = wer @@ -551,9 +547,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/decode.py index ce45a4beb..e497787d3 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/decode.py @@ -593,18 +593,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -614,9 +610,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/decode.py index aebe2b94b..e7616fbc5 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/decode.py @@ -568,18 +568,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -589,9 +585,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/streaming_decode.py b/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/streaming_decode.py index 7a349ecb2..c272ed641 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/streaming_decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7_streaming/streaming_decode.py @@ -409,18 +409,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -430,9 +426,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/pruned_transducer_stateless8/decode.py b/egs/librispeech/ASR/pruned_transducer_stateless8/decode.py index e61367134..7b651a632 100755 --- a/egs/librispeech/ASR/pruned_transducer_stateless8/decode.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless8/decode.py @@ -594,18 +594,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -615,9 +611,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/transducer/decode.py b/egs/librispeech/ASR/transducer/decode.py index 804713a20..8d379d1fa 100755 --- a/egs/librispeech/ASR/transducer/decode.py +++ b/egs/librispeech/ASR/transducer/decode.py @@ -325,18 +325,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -346,9 +342,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/transducer_lstm/decode.py b/egs/librispeech/ASR/transducer_lstm/decode.py index 9511ca6d7..806b68f40 100755 --- a/egs/librispeech/ASR/transducer_lstm/decode.py +++ b/egs/librispeech/ASR/transducer_lstm/decode.py @@ -322,18 +322,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -343,9 +339,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/transducer_stateless/decode.py b/egs/librispeech/ASR/transducer_stateless/decode.py index 643238f1b..42125e19f 100755 --- a/egs/librispeech/ASR/transducer_stateless/decode.py +++ b/egs/librispeech/ASR/transducer_stateless/decode.py @@ -379,18 +379,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -400,9 +396,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/transducer_stateless2/decode.py b/egs/librispeech/ASR/transducer_stateless2/decode.py index 9a6363629..b05fe2a4d 100755 --- a/egs/librispeech/ASR/transducer_stateless2/decode.py +++ b/egs/librispeech/ASR/transducer_stateless2/decode.py @@ -379,18 +379,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -400,9 +396,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/transducer_stateless_multi_datasets/decode.py b/egs/librispeech/ASR/transducer_stateless_multi_datasets/decode.py index 56ad558c6..5570b30ae 100755 --- a/egs/librispeech/ASR/transducer_stateless_multi_datasets/decode.py +++ b/egs/librispeech/ASR/transducer_stateless_multi_datasets/decode.py @@ -380,18 +380,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -401,9 +397,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/librispeech/ASR/zipformer_mmi/decode.py b/egs/librispeech/ASR/zipformer_mmi/decode.py index 7d0ea78bb..33c0bf199 100755 --- a/egs/librispeech/ASR/zipformer_mmi/decode.py +++ b/egs/librispeech/ASR/zipformer_mmi/decode.py @@ -471,18 +471,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats(f, f"{test_set_name}-{key}", results) test_set_wers[key] = wer @@ -490,9 +486,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/mgb2/ASR/pruned_transducer_stateless5/decode.py b/egs/mgb2/ASR/pruned_transducer_stateless5/decode.py index 1463f8f67..72338bade 100755 --- a/egs/mgb2/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/mgb2/ASR/pruned_transducer_stateless5/decode.py @@ -410,17 +410,13 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -430,9 +426,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/spgispeech/ASR/pruned_transducer_stateless2/decode.py b/egs/spgispeech/ASR/pruned_transducer_stateless2/decode.py index 219c96d60..4434aae62 100755 --- a/egs/spgispeech/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/spgispeech/ASR/pruned_transducer_stateless2/decode.py @@ -391,18 +391,14 @@ def save_results( test_set_wers = dict() test_set_cers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - wers_filename = ( - params.res_dir / f"wers-{test_set_name}-{key}-{params.suffix}.txt" - ) + wers_filename = params.res_dir / f"wers-{test_set_name}-{params.suffix}.txt" with open(wers_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -413,9 +409,7 @@ def save_results( results_char = [] for res in results: results_char.append((res[0], list("".join(res[1])), list("".join(res[2])))) - cers_filename = ( - params.res_dir / f"cers-{test_set_name}-{key}-{params.suffix}.txt" - ) + cers_filename = params.res_dir / f"cers-{test_set_name}-{params.suffix}.txt" with open(cers_filename, "w") as f: cer = write_error_stats( f, f"{test_set_name}-{key}", results_char, enable_log=True @@ -426,9 +420,7 @@ def save_results( test_set_wers = {k: v for k, v in sorted(test_set_wers.items(), key=lambda x: x[1])} test_set_cers = {k: v for k, v in sorted(test_set_cers.items(), key=lambda x: x[1])} - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER\tCER", file=f) for key in test_set_wers: diff --git a/egs/tal_csasr/ASR/pruned_transducer_stateless5/decode.py b/egs/tal_csasr/ASR/pruned_transducer_stateless5/decode.py index bf91fef7e..3bfb832fb 100755 --- a/egs/tal_csasr/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/tal_csasr/ASR/pruned_transducer_stateless5/decode.py @@ -509,18 +509,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -530,9 +526,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/tedlium3/ASR/pruned_transducer_stateless/decode.py b/egs/tedlium3/ASR/pruned_transducer_stateless/decode.py index 38f2ae83c..abba9d403 100755 --- a/egs/tedlium3/ASR/pruned_transducer_stateless/decode.py +++ b/egs/tedlium3/ASR/pruned_transducer_stateless/decode.py @@ -379,18 +379,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -400,9 +396,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/tedlium3/ASR/transducer_stateless/decode.py b/egs/tedlium3/ASR/transducer_stateless/decode.py index 01f08ce59..fb0e3116b 100755 --- a/egs/tedlium3/ASR/transducer_stateless/decode.py +++ b/egs/tedlium3/ASR/transducer_stateless/decode.py @@ -354,18 +354,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -375,9 +371,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/wenetspeech/ASR/pruned_transducer_stateless2/decode.py b/egs/wenetspeech/ASR/pruned_transducer_stateless2/decode.py index 04602ea2e..823b33ae5 100755 --- a/egs/wenetspeech/ASR/pruned_transducer_stateless2/decode.py +++ b/egs/wenetspeech/ASR/pruned_transducer_stateless2/decode.py @@ -516,18 +516,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -537,9 +533,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/wenetspeech/ASR/pruned_transducer_stateless5/decode.py b/egs/wenetspeech/ASR/pruned_transducer_stateless5/decode.py index 7bd1177bd..32d5738b1 100755 --- a/egs/wenetspeech/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/wenetspeech/ASR/pruned_transducer_stateless5/decode.py @@ -489,18 +489,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -510,9 +506,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/wenetspeech/ASR/pruned_transducer_stateless5/streaming_decode.py b/egs/wenetspeech/ASR/pruned_transducer_stateless5/streaming_decode.py index c7863415b..3a4dc3cb8 100644 --- a/egs/wenetspeech/ASR/pruned_transducer_stateless5/streaming_decode.py +++ b/egs/wenetspeech/ASR/pruned_transducer_stateless5/streaming_decode.py @@ -466,9 +466,7 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" # sort results so we can easily compare the difference between two # recognition results results = sorted(results) @@ -477,9 +475,7 @@ def save_results( # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -489,9 +485,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/xbmu_amdo31/ASR/pruned_transducer_stateless5/decode.py b/egs/xbmu_amdo31/ASR/pruned_transducer_stateless5/decode.py index 6a67e26f8..b77f734e3 100755 --- a/egs/xbmu_amdo31/ASR/pruned_transducer_stateless5/decode.py +++ b/egs/xbmu_amdo31/ASR/pruned_transducer_stateless5/decode.py @@ -701,18 +701,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -722,9 +718,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: diff --git a/egs/xbmu_amdo31/ASR/pruned_transducer_stateless7/decode.py b/egs/xbmu_amdo31/ASR/pruned_transducer_stateless7/decode.py index ace792e13..e334e690a 100755 --- a/egs/xbmu_amdo31/ASR/pruned_transducer_stateless7/decode.py +++ b/egs/xbmu_amdo31/ASR/pruned_transducer_stateless7/decode.py @@ -593,18 +593,14 @@ def save_results( ): test_set_wers = dict() for key, results in results_dict.items(): - recog_path = ( - params.res_dir / f"recogs-{test_set_name}-{key}-{params.suffix}.txt" - ) + recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt" results = sorted(results) store_transcripts(filename=recog_path, texts=results) logging.info(f"The transcripts are stored in {recog_path}") # The following prints out WERs, per-word error statistics and aligned # ref/hyp pairs. - errs_filename = ( - params.res_dir / f"errs-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt" with open(errs_filename, "w") as f: wer = write_error_stats( f, f"{test_set_name}-{key}", results, enable_log=True @@ -614,9 +610,7 @@ def save_results( logging.info("Wrote detailed error stats to {}".format(errs_filename)) test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1]) - errs_info = ( - params.res_dir / f"wer-summary-{test_set_name}-{key}-{params.suffix}.txt" - ) + errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt" with open(errs_info, "w") as f: print("settings\tWER", file=f) for key, val in test_set_wers: