Add CXX API examples for ten-vad. (#2380)

csukuangfj · web-flow · commit f2df11672606 · 2025-07-12T12:36:31.000+08:00
diff --git a/.github/workflows/cxx-api.yaml b/.github/workflows/cxx-api.yaml
@@ -196,7 +196,7 @@ jobs:
 
           rm $name
 
-      - name: Test VAD
+      - name: Test silero-vad
         shell: bash
         run: |
           name=vad-cxx-api
@@ -223,17 +223,60 @@ jobs:
 
           ./$name
 
-          mkdir vad-test
-          cp -v lei-jun-test*.wav vad-test
+          mkdir vad-test-silero-vad
+          cp -v lei-jun-test*.wav vad-test-silero-vad
 
-          ls -lh vad-test
+          ls -lh vad-test-silero-vad
 
           rm $name
+          rm -fv *.onnx
+          rm -fv *.wav
 
       - uses: actions/upload-artifact@v4
         with:
-          name: vad-test-wavs-cxx-${{ matrix.os }}
-          path: ./vad-test/*.wav
+          name: silero-vad-test-wavs-cxx-${{ matrix.os }}
+          path: ./vad-test-silero-vad/*.wav
+
+      - name: Test ten-vad
+        shell: bash
+        run: |
+          name=vad-cxx-api
+          g++ -std=c++17 -o $name ./cxx-api-examples/$name.cc \
+            -I ./build/install/include \
+            -L ./build/install/lib/ \
+            -l sherpa-onnx-cxx-api \
+            -l sherpa-onnx-c-api \
+            -l onnxruntime
+
+          ls -lh $name
+
+          export LD_LIBRARY_PATH=$PWD/build/install/lib:$LD_LIBRARY_PATH
+          export DYLD_LIBRARY_PATH=$PWD/build/install/lib:$DYLD_LIBRARY_PATH
+
+          if [[ ${{ matrix.os }} == ubuntu-latest || ${{ matrix.os }} == ubuntu-22.04-arm ]]; then
+            ldd ./$name
+            echo "----"
+            readelf -d ./$name
+          fi
+
+          curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
+          curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
+
+          ./$name
+
+          mkdir vad-test-ten-vad
+          cp -v lei-jun-test*.wav vad-test-ten-vad
+
+          ls -lh vad-test-ten-vad
+
+          rm $name
+          rm -fv *.onnx
+          rm -rf *.wav
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: ten-vad-test-wavs-cxx-${{ matrix.os }}
+          path: ./vad-test-ten-vad/*.wav
 
       - name: Test Speech Enhancement (GTCRN)
         shell: bash
diff --git a/cxx-api-examples/vad-cxx-api.cc b/cxx-api-examples/vad-cxx-api.cc
@@ -6,7 +6,12 @@
 // This file demonstrates how to use VAD to remove silences from a file
 // clang-format off
 //
-// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+// To use silero-vad:
+//  wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+//
+// To use ten-vad:
+//  wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
+//
 // wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
 //
 // clang-format on
@@ -19,14 +24,45 @@ int32_t main() {
   using namespace sherpa_onnx::cxx;  // NOLINT
 
   std::string wave_filename = "./lei-jun-test.wav";
-  std::string vad_filename = "./silero_vad.onnx";
+  if (!FileExists(wave_filename)) {
+    fprintf(stderr, "Please download %s\n", wave_filename.c_str());
+    return -1;
+  }
+
+  std::string vad_filename;
+  bool use_silero_vad = false;
+  bool use_ten_vad = false;
+
+  if (FileExists("./silero_vad.onnx")) {
+    printf("Use silero-vad\n");
+    vad_filename = "./silero_vad.onnx";
+    use_silero_vad = true;
+  } else if (FileExists("./ten-vad.onnx")) {
+    printf("Use ten-vad\n");
+    vad_filename = "./ten-vad.onnx";
+    use_ten_vad = true;
+  } else {
+    fprintf(stderr, "Please provide either silero_vad.onnx or ten-vad.onnx\n");
+    return -1;
+  }
 
   VadModelConfig config;
-  config.silero_vad.model = vad_filename;
-  config.silero_vad.threshold = 0.1;
-  config.silero_vad.min_silence_duration = 0.5;
-  config.silero_vad.min_speech_duration = 0.25;
-  config.silero_vad.max_speech_duration = 20;
+  if (use_silero_vad) {
+    config.silero_vad.model = vad_filename;
+    config.silero_vad.threshold = 0.3;
+    config.silero_vad.min_silence_duration = 0.5;
+    config.silero_vad.min_speech_duration = 0.25;
+    config.silero_vad.max_speech_duration = 20;
+    config.silero_vad.window_size = 512;
+  } else if (use_ten_vad) {
+    config.ten_vad.model = vad_filename;
+    config.ten_vad.threshold = 0.3;
+    config.ten_vad.min_silence_duration = 0.5;
+    config.ten_vad.min_speech_duration = 0.25;
+    config.ten_vad.max_speech_duration = 20;
+    config.ten_vad.window_size = 256;
+  }
+
   config.sample_rate = 16000;
   config.debug = true;
 
@@ -43,7 +79,8 @@ int32_t main() {
   }
   bool is_eof = false;
   int32_t i = 0;
-  int32_t window_size = config.silero_vad.window_size;
+  int32_t window_size = use_silero_vad ? config.silero_vad.window_size
+                                       : config.ten_vad.window_size;
 
   int32_t sample_rate = config.sample_rate;
 
diff --git a/sherpa-onnx/c-api/cxx-api.h b/sherpa-onnx/c-api/cxx-api.h
@@ -649,10 +649,10 @@ class SHERPA_ONNX_API LinearResampler
   explicit LinearResampler(const SherpaOnnxLinearResampler *p);
 };
 
-std::string GetVersionStr();
-std::string GetGitSha1();
-std::string GetGitDate();
-bool FileExists(const std::string &filename);
+SHERPA_ONNX_API std::string GetVersionStr();
+SHERPA_ONNX_API std::string GetGitSha1();
+SHERPA_ONNX_API std::string GetGitDate();
+SHERPA_ONNX_API bool FileExists(const std::string &filename);
 
 }  // namespace sherpa_onnx::cxx
 
