Add C++ runtime for Tele-AI/TeleSpeech-ASR (#970)

.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