Support onnxruntime 1.18.0

.github/scripts/test-nodejs-addon-npm.sh

@@ -121,9 +121,9 @@ rm sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k.tar.bz2
 node ./test_asr_non_streaming_nemo_ctc.js
 rm -rf sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k
 
-curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+tar xvf sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+rm sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
 
 node ./test_asr_non_streaming_paraformer.js
 
@@ -134,7 +134,7 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/i
 
 node ./test_asr_non_streaming_paraformer_itn.js
 
-rm -rf sherpa-onnx-paraformer-zh-2023-03-28
+rm -rf sherpa-onnx-paraformer-zh-2023-09-14
 
 echo "----------tts----------"
 

.github/scripts/test-nodejs-npm.sh

@@ -11,16 +11,16 @@ ls -lh node_modules
 
 # offline asr
 
-curl -LS -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+curl -LS -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
 ls -lh
-tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+tar xvf sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+rm sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
 
 rm -f itn*
 curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn-zh-number.wav
 curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn_zh_number.fst
 node ./test-offline-paraformer-itn.js
-rm -rf sherpa-onnx-paraformer-zh-2023-03-28
+rm -rf sherpa-onnx-paraformer-zh-2023-09-14
 
 curl -LS -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-ctc-en-conformer-small.tar.bz2
 ls -lh
@@ -29,12 +29,12 @@ rm sherpa-onnx-nemo-ctc-en-conformer-small.tar.bz2
 node ./test-offline-nemo-ctc.js
 rm -rf sherpa-onnx-nemo-ctc-en-conformer-small
 
-curl -LS -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+curl -LS -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
 ls -lh
-tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+tar xvf sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+rm sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
 node ./test-offline-paraformer.js
-rm -rf sherpa-onnx-paraformer-zh-2023-03-28
+rm -rf sherpa-onnx-paraformer-zh-2023-09-14
 
 curl -LS -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-zipformer-en-2023-06-26.tar.bz2
 ls -lh

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

@@ -15,6 +15,49 @@ echo "PATH: $PATH"
 
 which $EXE
 
+if false; then
+  # It has problems with onnxruntime 1.18
+  log "------------------------------------------------------------"
+  log "Run Wenet models"
+  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 ${wenet_models[@]}; do
+    repo_url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/$name.tar.bz2
+    log "Start testing ${repo_url}"
+    repo=$name
+    log "Download pretrained model and test-data from $repo_url"
+    curl -SL -O $repo_url
+    tar xvf $name.tar.bz2
+    rm $name.tar.bz2
+
+    log "test float32 models"
+    time $EXE \
+      --tokens=$repo/tokens.txt \
+      --wenet-ctc-model=$repo/model.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.int8.onnx \
+      $repo/test_wavs/0.wav \
+      $repo/test_wavs/1.wav \
+      $repo/test_wavs/8k.wav
+
+    rm -rf $repo
+  done
+fi
+
+
 log "test offline TeleSpeech CTC"
 url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04.tar.bz2
 name=$(basename $url)
@@ -145,44 +188,7 @@ 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-wenetspeech
-# sherpa-onnx-zh-wenet-multi-cn
-sherpa-onnx-en-wenet-librispeech
-# sherpa-onnx-en-wenet-gigaspeech
-)
-for name in ${wenet_models[@]}; do
-  repo_url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/$name.tar.bz2
-  log "Start testing ${repo_url}"
-  repo=$name
-  log "Download pretrained model and test-data from $repo_url"
-  curl -SL -O $repo_url
-  tar xvf $name.tar.bz2
-  rm $name.tar.bz2
-
-  log "test float32 models"
-  time $EXE \
-    --tokens=$repo/tokens.txt \
-    --wenet-ctc-model=$repo/model.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.int8.onnx \
-    $repo/test_wavs/0.wav \
-    $repo/test_wavs/1.wav \
-    $repo/test_wavs/8k.wav
-
-  rm -rf $repo
-done
 
 log "------------------------------------------------------------"
 log "Run tdnn yesno (Hebrew)"

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

@@ -184,35 +184,13 @@ rm -rf $repo
 log "------------------------------------------------------------"
 log "Run Paraformer (Chinese)"
 log "------------------------------------------------------------"
-
-repo_url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-curl -SL -O $repo_url
-tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-repo=sherpa-onnx-paraformer-zh-2023-03-28
-log "Start testing ${repo_url}"
-
-time $EXE \
-  --tokens=$repo/tokens.txt \
-  --paraformer=$repo/model.onnx \
-  --num-threads=2 \
-  --decoding-method=greedy_search \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav \
-  $repo/test_wavs/8k.wav
-
-time $EXE \
-  --tokens=$repo/tokens.txt \
-  --paraformer=$repo/model.int8.onnx \
-  --num-threads=2 \
-  --decoding-method=greedy_search \
-  $repo/test_wavs/0.wav \
-  $repo/test_wavs/1.wav \
-  $repo/test_wavs/2.wav \
-  $repo/test_wavs/8k.wav
-
-rm -rf $repo
+# For onnxruntime 1.18.0, sherpa-onnx-paraformer-zh-2023-03-28 throws the following error
+# libc++abi: terminating with uncaught exception of type Ort::Exception: Node (Loop_5471)
+# Op (Loop) [TypeInferenceError] Graph attribute inferencing failed: Node (Concat_5490)
+# Op (Concat) [ShapeInferenceError] All inputs to Concat must have same rank. Input 1 has rank 2 != 1
+#
+# See https://github.com/microsoft/onnxruntime/issues/8115
+# We need to re-export this model using a recent version of onnxruntime and onnx
 
 log "------------------------------------------------------------"
 log "Run Paraformer (Chinese) with timestamps"

.github/scripts/test-python.sh

@@ -312,13 +312,13 @@ log "Test non-streaming paraformer models"
 if [[ x$OS != x'windows-latest' ]]; then
   echo "OS: $OS"
   pushd $dir
-  repo_url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+  repo_url=https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
   curl -SL -O $repo_url
-  tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
-  rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+  tar xvf sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
+  rm sherpa-onnx-paraformer-zh-2023-09-14.tar.bz2
 
   log "Start testing ${repo_url}"
-  repo=$dir/sherpa-onnx-paraformer-zh-2023-03-28
+  repo=$dir/sherpa-onnx-paraformer-zh-2023-09-14
 
   ls -lh $repo
   popd
@@ -348,7 +348,7 @@ if [[ x$OS != x'windows-latest' ]]; then
 
   python3 ./python-api-examples/inverse-text-normalization-offline-asr.py
 
-  rm -rfv sherpa-onnx-paraformer-zh-2023-03-28
+  rm -rfv sherpa-onnx-paraformer-zh-2023-09-14
 
   rm -rf $repo
 fi

.github/workflows/arm-linux-gnueabihf.yaml

@@ -172,7 +172,6 @@ jobs:
           if [[ $lib_type == "shared" ]]; then
             cp -a build-arm-linux-gnueabihf/install/lib $dst/
             rm -v $dst/lib/libasound.so
-            rm -v $dst/lib/libonnxruntime.so
           fi
 
           tree $dst

.github/workflows/export-wenet-to-onnx.yaml

@@ -73,7 +73,14 @@ jobs:
             git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-zh-wenet-aishell main || true
 
             cd ..
-            rm -rf huggingface
+
+            rm -rf huggingface/.git
+            dst=sherpa-onnx-zh-wenet-aishell
+
+            mv huggingface $dst
+
+            tar cjvf $dst.tar.bz2 $dst
+            rm -rf $dst
 
       - name: Publish to huggingface (aishell2)
         env:
@@ -115,7 +122,14 @@ jobs:
             git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-zh-wenet-aishell2 main || true
 
             cd ..
-            rm -rf huggingface
+
+            rm -rf huggingface/.git
+            dst=sherpa-onnx-zh-wenet-aishell2
+
+            mv huggingface $dst
+
+            tar cjvf $dst.tar.bz2 $dst
+            rm -rf $dst
 
       - name: Publish to huggingface (multi_cn)
         env:
@@ -157,7 +171,14 @@ jobs:
             git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-zh-wenet-multi-cn main || true
 
             cd ..
-            rm -rf huggingface
+
+            rm -rf huggingface/.git
+            dst=sherpa-onnx-zh-wenet-multi-cn
+
+            mv huggingface $dst
+
+            tar cjvf $dst.tar.bz2 $dst
+            rm -rf $dst
 
       - name: Publish to huggingface (wenetspeech)
         env:
@@ -199,7 +220,14 @@ jobs:
             git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-zh-wenet-wenetspeech main || true
 
             cd ..
-            rm -rf huggingface
+
+            rm -rf huggingface/.git
+            dst=sherpa-onnx-zh-wenet-wenetspeech
+
+            mv huggingface $dst
+
+            tar cjvf $dst.tar.bz2 $dst
+            rm -rf $dst
 
       - name: Publish to huggingface (librispeech)
         env:
@@ -242,7 +270,14 @@ jobs:
             git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-en-wenet-librispeech main || true
 
             cd ..
-            rm -rf huggingface
+
+            rm -rf huggingface/.git
+            dst=sherpa-onnx-en-wenet-librispeech
+
+            mv huggingface $dst
+
+            tar cjvf $dst.tar.bz2 $dst
+            rm -rf $dst
 
       - name: Publish to huggingface (gigaspeech)
         env:
@@ -285,4 +320,21 @@ jobs:
             git push https://csukuangfj:$HF_TOKEN@huggingface.co/csukuangfj/sherpa-onnx-en-wenet-gigaspeech main || true
 
             cd ..
-            rm -rf huggingface
+
+            rm -rf huggingface/.git
+            dst=sherpa-onnx-en-wenet-gigaspeech
+
+            mv huggingface $dst
+
+            tar cjvf $dst.tar.bz2 $dst
+            rm -rf $dst
+
+      - name: Release
+        uses: svenstaro/upload-release-action@v2
+        with:
+          file_glob: true
+          file: ./*.tar.bz2
+          overwrite: true
+          repo_name: k2-fsa/sherpa-onnx
+          repo_token: ${{ secrets.UPLOAD_GH_SHERPA_ONNX_TOKEN }}
+          tag: asr-models

.github/workflows/release-dart-package.yaml

@@ -85,7 +85,6 @@ jobs:
               make install
 
               ls -lh ./install/lib
-              rm -v ./install/lib/libonnxruntime.so
 
       - name: Copy pre-built libs
         shell: bash

.github/workflows/test-go-package.yaml

@@ -269,7 +269,7 @@ jobs:
 
           echo "Test paraformer"
           ./run-paraformer.sh
-          rm -rf sherpa-onnx-paraformer-zh-2023-03-28
+          rm -rf sherpa-onnx-paraformer-zh-2023-09-14
 
           echo "Test NeMo CTC"
           ./run-nemo-ctc.sh
@@ -306,7 +306,7 @@ jobs:
 
           echo "Test paraformer"
           ./run-paraformer.sh
-          rm -rf sherpa-onnx-paraformer-zh-2023-03-28
+          rm -rf sherpa-onnx-paraformer-zh-2023-09-14
 
           echo "Test NeMo CTC"
           ./run-nemo-ctc.sh
@@ -351,7 +351,7 @@ jobs:
 
           echo "Test paraformer"
           ./run-paraformer.sh
-          rm -rf sherpa-onnx-paraformer-zh-2023-03-28
+          rm -rf sherpa-onnx-paraformer-zh-2023-09-14
 
           echo "Test NeMo CTC"
           ./run-nemo-ctc.sh