Support streaming conformer CTC models from wenet (#427)

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

@@ -0,0 +1,55 @@
+#!/usr/bin/env bash
+
+set -e
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+echo "EXE is $EXE"
+echo "PATH: $PATH"
+
+which $EXE
+
+log "------------------------------------------------------------"
+log "Run streaming Conformer CTC from WeNet"
+log "------------------------------------------------------------"
+wenet_models=(
+sherpa-onnx-zh-wenet-aishell
+sherpa-onnx-zh-wenet-aishell2
+sherpa-onnx-zh-wenet-wenetspeech
+sherpa-onnx-zh-wenet-multi-cn
+sherpa-onnx-en-wenet-librispeech
+sherpa-onnx-en-wenet-gigaspeech
+)
+for name in ${wenet_models[@]}; do
+  repo_url=https://huggingface.co/csukuangfj/$name
+  log "Start testing ${repo_url}"
+  repo=$(basename $repo_url)
+  log "Download pretrained model and test-data from $repo_url"
+  GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+  pushd $repo
+  git lfs pull --include "*.onnx"
+  ls -lh *.onnx
+  popd
+
+  log "test float32 models"
+  time $EXE \
+    --tokens=$repo/tokens.txt \
+    --wenet-ctc-model=$repo/model-streaming.onnx \
+    $repo/test_wavs/0.wav \
+    $repo/test_wavs/1.wav \
+    $repo/test_wavs/8k.wav
+
+  log "test int8 models"
+  time $EXE \
+    --tokens=$repo/tokens.txt \
+    --wenet-ctc-model=$repo/model-streaming.int8.onnx \
+    $repo/test_wavs/0.wav \
+    $repo/test_wavs/1.wav \
+    $repo/test_wavs/8k.wav
+
+  rm -rf $repo
+done

.github/workflows/linux-gpu.yaml

@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -27,6 +28,8 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -88,6 +91,14 @@ jobs:
           file build/bin/sherpa-onnx
           readelf -d 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 TTS
         shell: bash
         run: |

.github/workflows/linux.yaml

@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -27,6 +28,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -89,6 +91,14 @@ jobs:
           file build/bin/sherpa-onnx
           readelf -d 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 CTC
         shell: bash
         run: |

.github/workflows/macos.yaml

@@ -13,6 +13,7 @@ on:
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -26,6 +27,7 @@ on:
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -96,6 +98,15 @@ 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 TTS
         shell: bash
         run: |

.github/workflows/windows-x64-cuda.yaml

@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -25,6 +26,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -66,6 +68,14 @@ jobs:
 
           ls -lh ./bin/Release/sherpa-onnx.exe
 
+      - name: Test online CTC
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin/Release:$PATH
+          export EXE=sherpa-onnx.exe
+
+          .github/scripts/test-online-ctc.sh
+
       - name: Test offline TTS
         shell: bash
         run: |

.github/workflows/windows-x64.yaml

@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -25,6 +26,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
@@ -67,6 +69,14 @@ jobs:
 
           ls -lh ./bin/Release/sherpa-onnx.exe
 
+      - name: Test online CTC
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin/Release:$PATH
+          export EXE=sherpa-onnx.exe
+
+          .github/scripts/test-online-ctc.sh
+
       - name: Test offline TTS
         shell: bash
         run: |

.github/workflows/windows-x86.yaml

@@ -13,6 +13,7 @@ on:
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -26,6 +27,7 @@ on:
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
       - '.github/scripts/test-offline-tts.sh'
+      - '.github/scripts/test-online-ctc.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -67,6 +69,14 @@ jobs:
 
           ls -lh ./bin/Release/sherpa-onnx.exe
 
+      - name: Test online CTC
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin/Release:$PATH
+          export EXE=sherpa-onnx.exe
+
+          .github/scripts/test-online-ctc.sh
+
       - name: Test offline TTS
         shell: bash
         run: |