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

scripts/apk/generate-vad-asr-apk-script.py

@@ -163,6 +163,22 @@ def get_models():
 
             ls -lh
 
+            popd
+            """,
+        ),
+        Model(
+            model_name="sherpa-onnx-telespeech-ctc-int8-zh-2024-06-04",
+            idx=11,
+            lang="zh",
+            short_name="telespeech",
+            cmd="""
+            pushd $model_name
+
+            rm -rfv test_wavs
+            rm test.py
+
+            ls -lh
+
             popd
             """,
         ),

scripts/dotnet/OfflineModelConfig.cs

@@ -25,6 +25,7 @@ namespace SherpaOnnx
             ModelType = "";
             ModelingUnit = "cjkchar";
             BpeVocab = "";
+            TeleSpeechCtc = "";
         }
         public OfflineTransducerModelConfig Transducer;
         public OfflineParaformerModelConfig Paraformer;
@@ -50,5 +51,8 @@ namespace SherpaOnnx
 
         [MarshalAs(UnmanagedType.LPStr)]
         public string BpeVocab;
+
+        [MarshalAs(UnmanagedType.LPStr)]
+        public string TeleSpeechCtc;
     }
 }

scripts/dotnet/run.sh

@@ -30,7 +30,7 @@ mkdir -p linux macos windows-x64 windows-x86
 linux_wheel_filename=sherpa_onnx-${SHERPA_ONNX_VERSION}-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
 linux_wheel=$src_dir/$linux_wheel_filename
 
-macos_wheel_filename=sherpa_onnx-${SHERPA_ONNX_VERSION}-cp38-cp38-macosx_11_0_x86_64.whl
+macos_wheel_filename=sherpa_onnx-${SHERPA_ONNX_VERSION}-cp38-cp38-macosx_11_0_universal2.whl
 macos_wheel=$src_dir/$macos_wheel_filename
 
 windows_x64_wheel_filename=sherpa_onnx-${SHERPA_ONNX_VERSION}-cp38-cp38-win_amd64.whl
@@ -61,7 +61,7 @@ if [ ! -f $src_dir/linux/libsherpa-onnx-core.so ]; then
 fi
 
 if [ ! -f $src_dir/macos/libsherpa-onnx-core.dylib ]; then
-  echo "---macOS x86_64---"
+  echo "--- macOS x86_64/arm64 universal2---"
   cd macos
   mkdir -p wheel
   cd wheel

scripts/go/_internal/non-streaming-decode-files/run-telespeech-ctc.sh

@@ -0,0 +1 @@
+../../../../go-api-examples/non-streaming-decode-files/run-telespeech-ctc.sh

scripts/go/sherpa_onnx.go

@@ -381,8 +381,9 @@ type OfflineModelConfig struct {
 	// Optional. Specify it for faster model initialization.
 	ModelType string
 
-	ModelingUnit string // Optional. cjkchar, bpe, cjkchar+bpe
-	BpeVocab     string // Optional.
+	ModelingUnit  string // Optional. cjkchar, bpe, cjkchar+bpe
+	BpeVocab      string // Optional.
+	TeleSpeechCtc string // Optional.
 }
 
 // Configuration for the offline/non-streaming recognizer.
@@ -477,6 +478,9 @@ func NewOfflineRecognizer(config *OfflineRecognizerConfig) *OfflineRecognizer {
 	c.model_config.bpe_vocab = C.CString(config.ModelConfig.BpeVocab)
 	defer C.free(unsafe.Pointer(c.model_config.bpe_vocab))
 
+	c.model_config.telespeech_ctc = C.CString(config.ModelConfig.TeleSpeechCtc)
+	defer C.free(unsafe.Pointer(c.model_config.telespeech_ctc))
+
 	c.lm_config.model = C.CString(config.LmConfig.Model)
 	defer C.free(unsafe.Pointer(c.lm_config.model))
 

scripts/node-addon-api/src/non-streaming-asr.cc

@@ -128,6 +128,7 @@ static SherpaOnnxOfflineModelConfig GetOfflineModelConfig(Napi::Object obj) {
   SHERPA_ONNX_ASSIGN_ATTR_STR(model_type, modelType);
   SHERPA_ONNX_ASSIGN_ATTR_STR(modeling_unit, modelingUnit);
   SHERPA_ONNX_ASSIGN_ATTR_STR(bpe_vocab, bpeVocab);
+  SHERPA_ONNX_ASSIGN_ATTR_STR(telespeech_ctc, teleSpeechCtc);
 
   return c;
 }
@@ -242,6 +243,10 @@ CreateOfflineRecognizerWrapper(const Napi::CallbackInfo &info) {
     delete[] c.model_config.bpe_vocab;
   }
 
+  if (c.model_config.telespeech_ctc) {
+    delete[] c.model_config.telespeech_ctc;
+  }
+
   if (c.lm_config.model) {
     delete[] c.lm_config.model;
   }