Add C++ runtime for Tele-AI/TeleSpeech-ASR

.github/scripts/test-dot-net.sh

@@ -2,7 +2,16 @@
 
 cd dotnet-examples/
 
-cd vad-non-streaming-asr-paraformer
+cd ./offline-decode-files
+./run-telespeech-ctc.sh
+./run-nemo-ctc.sh
+./run-paraformer.sh
+./run-zipformer.sh
+./run-hotwords.sh
+./run-whisper.sh
+./run-tdnn-yesno.sh
+
+cd ../vad-non-streaming-asr-paraformer
 ./run.sh
 
 cd ../offline-punctuation
@@ -22,14 +31,6 @@ cd ../online-decode-files
 ./run-transducer.sh
 ./run-paraformer.sh
 
-cd ../offline-decode-files
-./run-nemo-ctc.sh
-./run-paraformer.sh
-./run-zipformer.sh
-./run-hotwords.sh
-./run-whisper.sh
-./run-tdnn-yesno.sh
-
 cd ../offline-tts
 ./run-aishell3.sh
 ./run-piper.sh

.github/scripts/test-offline-ctc.sh

@@ -15,6 +15,39 @@ echo "PATH: $PATH"
 
 which $EXE
 
+log "test offline TeleSpeech CTC"
+url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04.tar.bz2
+name=$(basename $url)
+repo=$(basename -s .tar.bz2 $name)
+
+curl -SL -O $url
+tar xvf $name
+rm $name
+ls -lh $repo
+
+test_wavs=(
+3-sichuan.wav
+4-tianjin.wav
+5-henan.wav
+)
+for w in ${test_wavs[@]}; do
+  time $EXE \
+    --tokens=$repo/tokens.txt \
+    --telespeech-ctc=$repo/model.int8.onnx \
+    --debug=1 \
+    $repo/test_wavs/$w
+done
+
+time $EXE \
+  --tokens=$repo/tokens.txt \
+  --telespeech-ctc=$repo/model.int8.onnx \
+  --debug=1 \
+  $repo/test_wavs/3-sichuan.wav \
+  $repo/test_wavs/4-tianjin.wav \
+  $repo/test_wavs/5-henan.wav
+
+rm -rf $repo
+
 log "-----------------------------------------------------------------"
 log "Run Nemo fast conformer hybrid transducer ctc models (CTC branch)"
 log "-----------------------------------------------------------------"

.github/scripts/test-python.sh

@@ -10,6 +10,18 @@ log() {
 
 export GIT_CLONE_PROTECTION_ACTIVE=false
 
+log "test offline TeleSpeech CTC"
+url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04.tar.bz2
+name=$(basename $url)
+repo=$(basename -s .tar.bz2 $name)
+
+curl -SL -O $url
+tar xvf $name
+rm $name
+ls -lh $repo
+python3 ./python-api-examples/offline-telespeech-ctc-decode-files.py
+rm -rf $repo
+
 log "test online NeMo CTC"
 
 url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-streaming-fast-conformer-ctc-en-80ms.tar.bz2

.github/workflows/build-wheels-macos-arm64.yaml

@@ -82,7 +82,7 @@ jobs:
           TWINE_USERNAME: ${{ secrets.PYPI_USERNAME }}
           TWINE_PASSWORD: ${{ secrets.PYPI_PASSWORD }}
         run: |
-          python3 -m pip install --upgrade pip
-          python3 -m pip install wheel twine setuptools
+          python3 -m pip install --break-system-packages --upgrade pip
+          python3 -m pip install --break-system-packages wheel twine setuptools
 
           twine upload ./wheelhouse/*.whl

.github/workflows/build-wheels-macos-universal2.yaml

@@ -0,0 +1,90 @@
+name: build-wheels-macos-universal2
+
+on:
+  push:
+    branches:
+      - wheel
+    tags:
+      - '*'
+  workflow_dispatch:
+
+env:
+  SHERPA_ONNX_IS_IN_GITHUB_ACTIONS: 1
+
+concurrency:
+  group: build-wheels-macos-universal2-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  build_wheels_macos_universal2:
+    name: ${{ matrix.python-version }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [macos-latest]
+        python-version: ["cp38", "cp39", "cp310", "cp311", "cp312"]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Build wheels
+        uses: pypa/cibuildwheel@v2.15.0
+        env:
+          CIBW_BUILD: "${{ matrix.python-version}}-* "
+          CIBW_ENVIRONMENT: SHERPA_ONNX_CMAKE_ARGS="-DCMAKE_OSX_ARCHITECTURES='arm64;x86_64'"
+          CIBW_ARCHS: "universal2"
+          CIBW_BUILD_VERBOSITY: 3
+
+          #  Don't repair macOS wheels
+          CIBW_REPAIR_WHEEL_COMMAND_MACOS: ""
+
+      - name: Display wheels
+        shell: bash
+        run: |
+          ls -lh ./wheelhouse/
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: wheel-${{ matrix.python-version }}
+          path: ./wheelhouse/*.whl
+
+      - name: Publish to huggingface
+        if: matrix.python-version == 'cp38'
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
+        uses: nick-fields/retry@v3
+        with:
+          max_attempts: 20
+          timeout_seconds: 200
+          shell: bash
+          command: |
+            git config --global user.email "csukuangfj@gmail.com"
+            git config --global user.name "Fangjun Kuang"
+
+            rm -rf huggingface
+            export GIT_LFS_SKIP_SMUDGE=1
+            export GIT_CLONE_PROTECTION_ACTIVE=false
+
+            git clone https://huggingface.co/csukuangfj/sherpa-onnx-wheels huggingface
+            cd huggingface
+            git fetch
+            git pull
+            git merge -m "merge remote" --ff origin main
+
+            cp -v ../wheelhouse/*.whl .
+
+            git status
+            git add .
+            git commit -m "add more wheels"
+            git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-wheels main
+
+      - name: Publish wheels to PyPI
+        env:
+          TWINE_USERNAME: ${{ secrets.PYPI_USERNAME }}
+          TWINE_PASSWORD: ${{ secrets.PYPI_PASSWORD }}
+        run: |
+          python3 -m pip install --break-system-packages --upgrade pip
+          python3 -m pip install --break-system-packages wheel twine setuptools
+
+          twine upload ./wheelhouse/*.whl

.github/workflows/build-wheels-macos-x64.yaml

@@ -99,7 +99,7 @@ jobs:
           TWINE_USERNAME: ${{ secrets.PYPI_USERNAME }}
           TWINE_PASSWORD: ${{ secrets.PYPI_PASSWORD }}
         run: |
-          python3 -m pip install --upgrade pip
-          python3 -m pip install wheel twine setuptools
+          python3 -m pip install --break-system-packages --upgrade pip
+          python3 -m pip install --break-system-packages wheel twine setuptools
 
           twine upload ./wheelhouse/*.whl

.github/workflows/export-telespeech-ctc.yaml

@@ -48,3 +48,49 @@ jobs:
           repo_name: k2-fsa/sherpa-onnx
           repo_token: ${{ secrets.UPLOAD_GH_SHERPA_ONNX_TOKEN }}
           tag: asr-models
+
+      - name: Publish float32 model to huggingface
+        shell: bash
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
+        run: |
+          src=scripts/tele-speech/sherpa-onnx-telespeech-ctc-zh-2024-06-04
+          git config --global user.email "csukuangfj@gmail.com"
+          git config --global user.name "Fangjun Kuang"
+
+          export GIT_CLONE_PROTECTION_ACTIVE=false
+
+          GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/csukuangfj/sherpa-onnx-telespeech-ctc-zh-2024-06-04 hf
+          cp -a $src/* hf/
+          cd hf
+          git lfs track "*.pdf"
+          git lfs track "*.onnx"
+          git add .
+          git commit -m 'add model files' || true
+          git status
+          ls -lh
+          git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-telespeech-ctc-zh-2024-06-04 main || true
+          rm -rf hf
+
+      - name: Publish int8 model to huggingface
+        shell: bash
+        env:
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
+        run: |
+          src=scripts/tele-speech/sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04
+          git config --global user.email "csukuangfj@gmail.com"
+          git config --global user.name "Fangjun Kuang"
+
+          export GIT_CLONE_PROTECTION_ACTIVE=false
+
+          rm -rf hf
+          GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/csukuangfj/sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04 hf
+          cp -a $src/* hf/
+          cd hf
+          git lfs track "*.pdf"
+          git lfs track "*.onnx"
+          git add .
+          git commit -m 'add model files' || true
+          git status
+          ls -lh
+          git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04 main || true

.github/workflows/linux.yaml

@@ -130,34 +130,34 @@ jobs:
           name: release-${{ matrix.build_type }}-with-shared-lib-${{ matrix.shared_lib }}-with-tts-${{ matrix.with_tts }}
           path: install/*
 
-      - name: Test online transducer
+      - name: Test offline CTC
         shell: bash
         run: |
           du -h -d1 .
           export PATH=$PWD/build/bin:$PATH
-          export EXE=sherpa-onnx
+          export EXE=sherpa-onnx-offline
 
-          .github/scripts/test-online-transducer.sh
+          .github/scripts/test-offline-ctc.sh
           du -h -d1 .
 
-      - name: Test online transducer (C API)
+      - name: Test online transducer
         shell: bash
         run: |
           du -h -d1 .
           export PATH=$PWD/build/bin:$PATH
-          export EXE=decode-file-c-api
+          export EXE=sherpa-onnx
 
           .github/scripts/test-online-transducer.sh
           du -h -d1 .
 
-      - name: Test offline CTC
+      - name: Test online transducer (C API)
         shell: bash
         run: |
           du -h -d1 .
           export PATH=$PWD/build/bin:$PATH
-          export EXE=sherpa-onnx-offline
+          export EXE=decode-file-c-api
 
-          .github/scripts/test-offline-ctc.sh
+          .github/scripts/test-online-transducer.sh
           du -h -d1 .
 
       - name: Test spoken language identification (C++ API)

.github/workflows/macos.yaml

@@ -107,6 +107,14 @@ jobs:
           otool -L build/bin/sherpa-onnx
           otool -l build/bin/sherpa-onnx
 
+      - 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 transducer
         shell: bash
         run: |
@@ -192,13 +200,7 @@ 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 online transducer
         shell: bash

.github/workflows/swift.yaml

@@ -39,7 +39,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [macos-13]
+        os: [macos-latest, macos-14]
 
     steps:
       - uses: actions/checkout@v4

.github/workflows/test-go.yaml

@@ -30,14 +30,12 @@ concurrency:
 
 jobs:
   test-go:
-    name: ${{ matrix.os }} ${{matrix.arch }}
+    name: ${{ matrix.os }}
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        include:
-          - os: macos-latest
-            arch: amd64
+        os: [macos-latest, macos-14]
 
     steps:
       - uses: actions/checkout@v4
@@ -47,7 +45,7 @@ jobs:
       - name: ccache
         uses: hendrikmuhs/ccache-action@v1.2
         with:
-          key: ${{ matrix.os }}-${{ matrix.arch }}
+          key: ${{ matrix.os }}-go
 
       - uses: actions/setup-go@v5
         with:
@@ -109,8 +107,6 @@ jobs:
           go build
           ls -lh
 
-          git lfs install
-
           echo "Test vits-ljs"
           ./run-vits-ljs.sh
           rm -rf vits-ljs
@@ -144,7 +140,13 @@ jobs:
           go build
           ls -lh
 
-          git lfs install
+          echo "Test telespeech ctc"
+          ./run-telespeech-ctc.sh
+          rm -rf sherpa-onnx-telespeech-ctc-*
+
+          echo "Test transducer"
+          ./run-transducer.sh
+          rm -rf sherpa-onnx-zipformer-en-2023-06-26
 
           echo "Test transducer"
           ./run-transducer.sh

.github/workflows/test-piper-phonemize.yaml

@@ -57,7 +57,7 @@ jobs:
 
           mkdir build
           cd build
-          cmake -DCMAKE_VERBOSE_MAKEFILE=ON -D SHERPA_ONNX_ENABLE_TESTS=ON -D CMAKE_BUILD_TYPE=${{ matrix.build_type }} -D BUILD_SHARED_LIBS=${{ matrix.shared_lib }} -DCMAKE_INSTALL_PREFIX=./install ..
+          cmake -DSHERPA_ONNX_ENABLE_EPSEAK_NG_EXE=ON -DBUILD_ESPEAK_NG_EXE=ON -DCMAKE_VERBOSE_MAKEFILE=ON -D SHERPA_ONNX_ENABLE_TESTS=ON -D CMAKE_BUILD_TYPE=${{ matrix.build_type }} -D BUILD_SHARED_LIBS=${{ matrix.shared_lib }} -DCMAKE_INSTALL_PREFIX=./install ..
 
       - name: Build
         shell: bash

.gitignore

@@ -106,3 +106,4 @@ node_modules
 package-lock.json
 sherpa-onnx-nemo-*
 sherpa-onnx-vits-*
+sherpa-onnx-telespeech-ctc-*