Add C++ support for non-streaming NeMo fast conformer hybrid transducer ctc (the ctc branch) (#848)
This commit is contained in:
Fangjun Kuang
101
.github/scripts/test-offline-ctc.sh
vendored
101
.github/scripts/test-offline-ctc.sh
vendored
@@ -13,14 +13,111 @@ echo "PATH: $PATH"
|
||||
|
||||
which $EXE
|
||||
|
||||
log "-----------------------------------------------------------------"
|
||||
log "Run Nemo fast conformer hybrid transducer ctc models (CTC branch)"
|
||||
log "-----------------------------------------------------------------"
|
||||
|
||||
url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k.tar.bz2
|
||||
name=$(basename $url)
|
||||
curl -SL -O $url
|
||||
tar xvf $name
|
||||
rm $name
|
||||
repo=$(basename -s .tar.bz2 $name)
|
||||
ls -lh $repo
|
||||
|
||||
log "test $repo"
|
||||
test_wavs=(
|
||||
de-german.wav
|
||||
es-spanish.wav
|
||||
hr-croatian.wav
|
||||
po-polish.wav
|
||||
uk-ukrainian.wav
|
||||
en-english.wav
|
||||
fr-french.wav
|
||||
it-italian.wav
|
||||
ru-russian.wav
|
||||
)
|
||||
for w in ${test_wavs[@]}; do
|
||||
time $EXE \
|
||||
--tokens=$repo/tokens.txt \
|
||||
--nemo-ctc-model=$repo/model.onnx \
|
||||
--debug=1 \
|
||||
$repo/test_wavs/$w
|
||||
done
|
||||
|
||||
rm -rf $repo
|
||||
|
||||
url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-fast-conformer-ctc-en-24500.tar.bz2
|
||||
name=$(basename $url)
|
||||
curl -SL -O $url
|
||||
tar xvf $name
|
||||
rm $name
|
||||
repo=$(basename -s .tar.bz2 $name)
|
||||
ls -lh $repo
|
||||
|
||||
log "Test $repo"
|
||||
|
||||
time $EXE \
|
||||
--tokens=$repo/tokens.txt \
|
||||
--nemo-ctc-model=$repo/model.onnx \
|
||||
--debug=1 \
|
||||
$repo/test_wavs/en-english.wav
|
||||
|
||||
rm -rf $repo
|
||||
|
||||
url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-fast-conformer-ctc-es-1424.tar.bz2
|
||||
name=$(basename $url)
|
||||
curl -SL -O $url
|
||||
tar xvf $name
|
||||
rm $name
|
||||
repo=$(basename -s .tar.bz2 $name)
|
||||
ls -lh $repo
|
||||
|
||||
log "test $repo"
|
||||
|
||||
time $EXE \
|
||||
--tokens=$repo/tokens.txt \
|
||||
--nemo-ctc-model=$repo/model.onnx \
|
||||
--debug=1 \
|
||||
$repo/test_wavs/es-spanish.wav
|
||||
|
||||
rm -rf $repo
|
||||
|
||||
url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-fast-conformer-ctc-en-de-es-fr-14288.tar.bz2
|
||||
name=$(basename $url)
|
||||
curl -SL -O $url
|
||||
tar xvf $name
|
||||
rm $name
|
||||
repo=$(basename -s .tar.bz2 $name)
|
||||
ls -lh $repo
|
||||
|
||||
log "Test $repo"
|
||||
|
||||
test_wavs=(
|
||||
en-english.wav
|
||||
de-german.wav
|
||||
fr-french.wav
|
||||
es-spanish.wav
|
||||
)
|
||||
|
||||
for w in ${test_wavs[@]}; do
|
||||
time $EXE \
|
||||
--tokens=$repo/tokens.txt \
|
||||
--nemo-ctc-model=$repo/model.onnx \
|
||||
--debug=1 \
|
||||
$repo/test_wavs/$w
|
||||
done
|
||||
|
||||
rm -rf $repo
|
||||
|
||||
log "------------------------------------------------------------"
|
||||
log "Run Wenet models"
|
||||
log "------------------------------------------------------------"
|
||||
wenet_models=(
|
||||
sherpa-onnx-zh-wenet-aishell
|
||||
sherpa-onnx-zh-wenet-aishell2
|
||||
# sherpa-onnx-zh-wenet-aishell2
|
||||
# sherpa-onnx-zh-wenet-wenetspeech
|
||||
sherpa-onnx-zh-wenet-multi-cn
|
||||
# sherpa-onnx-zh-wenet-multi-cn
|
||||
sherpa-onnx-en-wenet-librispeech
|
||||
# sherpa-onnx-en-wenet-gigaspeech
|
||||
)
|
||||
|
||||
@@ -62,6 +62,11 @@ for wav in ${waves[@]}; do
|
||||
ls -lh *.wav
|
||||
done
|
||||
|
||||
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/spoken-language-identification-test-wavs.tar.bz2
|
||||
tar xvf spoken-language-identification-test-wavs.tar.bz2
|
||||
rm spoken-language-identification-test-wavs.tar.bz2
|
||||
data=spoken-language-identification-test-wavs
|
||||
|
||||
for name in ${names[@]}; do
|
||||
log "------------------------------------------------------------"
|
||||
log "Run $name"
|
||||
@@ -85,14 +90,14 @@ for name in ${names[@]}; do
|
||||
time $EXE \
|
||||
--whisper-encoder=$repo/${name}-encoder.onnx \
|
||||
--whisper-decoder=$repo/${name}-decoder.onnx \
|
||||
$wav
|
||||
$data/$wav
|
||||
|
||||
log "test int8 onnx"
|
||||
|
||||
time $EXE \
|
||||
--whisper-encoder=$repo/${name}-encoder.int8.onnx \
|
||||
--whisper-decoder=$repo/${name}-decoder.int8.onnx \
|
||||
$wav
|
||||
$data/$wav
|
||||
done
|
||||
rm -rf $repo
|
||||
done
|
||||
|
||||
36
.github/workflows/linux.yaml
vendored
36
.github/workflows/linux.yaml
vendored
@@ -128,13 +128,13 @@ jobs:
|
||||
name: release-${{ matrix.build_type }}-with-shared-lib-${{ matrix.shared_lib }}-with-tts-${{ matrix.with_tts }}
|
||||
path: install/*
|
||||
|
||||
- name: Test offline punctuation
|
||||
- name: Test spoken language identification (C++ API)
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx-offline-punctuation
|
||||
export EXE=sherpa-onnx-offline-language-identification
|
||||
|
||||
.github/scripts/test-offline-punctuation.sh
|
||||
.github/scripts/test-spoken-language-identification.sh
|
||||
|
||||
- name: Test C API
|
||||
shell: bash
|
||||
@@ -147,13 +147,13 @@ jobs:
|
||||
|
||||
.github/scripts/test-c-api.sh
|
||||
|
||||
- name: Test Audio tagging
|
||||
- name: Test offline CTC
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx-offline-audio-tagging
|
||||
export EXE=sherpa-onnx-offline
|
||||
|
||||
.github/scripts/test-audio-tagging.sh
|
||||
.github/scripts/test-offline-ctc.sh
|
||||
|
||||
- name: Test online CTC
|
||||
shell: bash
|
||||
@@ -163,14 +163,21 @@ jobs:
|
||||
|
||||
.github/scripts/test-online-ctc.sh
|
||||
|
||||
|
||||
- name: Test spoken language identification (C++ API)
|
||||
- name: Test offline punctuation
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx-offline-language-identification
|
||||
export EXE=sherpa-onnx-offline-punctuation
|
||||
|
||||
.github/scripts/test-spoken-language-identification.sh
|
||||
.github/scripts/test-offline-punctuation.sh
|
||||
|
||||
- name: Test Audio tagging
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx-offline-audio-tagging
|
||||
|
||||
.github/scripts/test-audio-tagging.sh
|
||||
|
||||
- name: Test transducer kws
|
||||
shell: bash
|
||||
@@ -180,7 +187,6 @@ jobs:
|
||||
|
||||
.github/scripts/test-kws.sh
|
||||
|
||||
|
||||
- name: Test offline Whisper
|
||||
if: matrix.build_type != 'Debug'
|
||||
shell: bash
|
||||
@@ -192,14 +198,6 @@ jobs:
|
||||
|
||||
.github/scripts/test-offline-whisper.sh
|
||||
|
||||
- name: Test offline CTC
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx-offline
|
||||
|
||||
.github/scripts/test-offline-ctc.sh
|
||||
|
||||
- name: Test offline TTS
|
||||
if: matrix.with_tts == 'ON'
|
||||
shell: bash
|
||||
|
||||
16
.github/workflows/macos.yaml
vendored
16
.github/workflows/macos.yaml
vendored
@@ -107,6 +107,14 @@ jobs:
|
||||
otool -L build/bin/sherpa-onnx
|
||||
otool -l build/bin/sherpa-onnx
|
||||
|
||||
- name: Test online CTC
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx
|
||||
|
||||
.github/scripts/test-online-ctc.sh
|
||||
|
||||
- name: Test offline punctuation
|
||||
shell: bash
|
||||
run: |
|
||||
@@ -150,14 +158,6 @@ jobs:
|
||||
|
||||
.github/scripts/test-kws.sh
|
||||
|
||||
- name: Test online CTC
|
||||
shell: bash
|
||||
run: |
|
||||
export PATH=$PWD/build/bin:$PATH
|
||||
export EXE=sherpa-onnx
|
||||
|
||||
.github/scripts/test-online-ctc.sh
|
||||
|
||||
- name: Test offline TTS
|
||||
if: matrix.with_tts == 'ON'
|
||||
shell: bash
|
||||
|
||||
Reference in New Issue
Block a user