Add Pascal API for ten-vad (#2388)

@@ -136,6 +136,27 @@ jobs:
             cp -v ../sherpa-onnx/pascal-api/*.pas ../pascal-api-examples/vad-with-non-streaming-asr
           fi
 
+      - name:  Run Pascal test (VAD test)
+        shell: bash
+        run: |
+          export PATH=/c/lazarus/fpc/3.2.2/bin/x86_64-win64:$PATH
+
+          cd ./pascal-api-examples
+
+          pushd vad
+          ./run-circular-buffer.sh
+          echo "---"
+
+          time ./run-remove-silence-ten-vad.sh
+          echo "---"
+
+          time ./run-remove-silence.sh
+          echo "---"
+
+          ls -lh
+
+          popd
+
       - name:  Run Speech Enhancement test (GTCRN)
         shell: bash
         run: |
@@ -298,24 +319,6 @@ jobs:
 
           popd
 
-      - name:  Run Pascal test (VAD test)
-        shell: bash
-        run: |
-          export PATH=/c/lazarus/fpc/3.2.2/bin/x86_64-win64:$PATH
-
-          cd ./pascal-api-examples
-
-          pushd vad
-          ./run-circular-buffer.sh
-          echo "---"
-
-          time ./run-remove-silence.sh
-          echo "---"
-
-          ls -lh
-
-          popd
-
       - name:  Run Pascal test (Read wav test)
         shell: bash
         run: |

 !run*.sh
 circular_buffer
 remove_silence
+remove_silence_ten_vad

 { Copyright (c)  2024  Xiaomi Corporation }
 {
 This file shows how to use the VAD API from sherpa-onnx
-to remove silences from a wave file.
+to remove silences from a wave file with silero-vad.
 }
 program main;
 

+{ Copyright (c)  2025  Xiaomi Corporation }
+{
+This file shows how to use the VAD API from sherpa-onnx
+to remove silences from a wave file with ten-vad.
+}
+program main;
+
+{$mode delphi}
+
+uses
+  sherpa_onnx,
+  SysUtils;
+
+var
+  Wave: TSherpaOnnxWave;
+
+  Config: TSherpaOnnxVadModelConfig;
+  Vad: TSherpaOnnxVoiceActivityDetector;
+  Offset: Integer;
+  WindowSize: Integer;
+  SpeechSegment: TSherpaOnnxSpeechSegment;
+
+  Start: Single;
+  Duration: Single;
+  SampleRate: Integer;
+
+  AllSpeechSegment: array of TSherpaOnnxSpeechSegment;
+  AllSamples: array of Single;
+  N: Integer;
+  I: Integer;
+begin
+  SampleRate := 16000; {Please don't change it unless you know the details}
+
+  Wave := SherpaOnnxReadWave('./lei-jun-test.wav');
+  if Wave.SampleRate <> SampleRate then
+    begin
+      WriteLn(Format('Expected sample rate: %d. Given: %d',
+        [SampleRate, Wave.SampleRate]));
+
+      Exit;
+    end;
+
+  WindowSize := 256; {Please don't change it unless you know the details}
+  Initialize(Config);
+
+  Config.TenVad.Model := './ten-vad.onnx';
+  Config.TenVad.MinSpeechDuration := 0.25;
+  Config.TenVad.MinSilenceDuration := 0.5;
+  Config.TenVad.Threshold := 0.25;
+  Config.TenVad.WindowSize := WindowSize;
+  Config.NumThreads:= 1;
+  Config.Debug:= True;
+  Config.Provider:= 'cpu';
+  Config.SampleRate := SampleRate;
+
+  Vad := TSherpaOnnxVoiceActivityDetector.Create(Config, 20);
+
+  AllSpeechSegment := nil;
+  AllSamples := nil;
+  Offset := 0;
+  while Offset + WindowSize <= Length(Wave.Samples) do
+    begin
+      Vad.AcceptWaveform(Wave.Samples, Offset, WindowSize);
+      Inc(Offset, WindowSize);
+
+      while not Vad.IsEmpty do
+        begin
+          SetLength(AllSpeechSegment, Length(AllSpeechSegment) + 1);
+
+          SpeechSegment := Vad.Front();
+          Vad.Pop();
+          AllSpeechSegment[Length(AllSpeechSegment)-1] := SpeechSegment;
+
+          Start := SpeechSegment.Start / SampleRate;
+          Duration := Length(SpeechSegment.Samples) / SampleRate;
+          WriteLn(Format('%.3f -- %.3f', [Start, Start + Duration]));
+        end;
+    end;
+
+  Vad.Flush;
+
+  while not Vad.IsEmpty do
+    begin
+      SetLength(AllSpeechSegment, Length(AllSpeechSegment) + 1);
+
+      SpeechSegment := Vad.Front();
+      Vad.Pop();
+      AllSpeechSegment[Length(AllSpeechSegment)-1] := SpeechSegment;
+
+      Start := SpeechSegment.Start / SampleRate;
+      Duration := Length(SpeechSegment.Samples) / SampleRate;
+      WriteLn(Format('%.3f -- %.3f', [Start, Start + Duration]));
+    end;
+
+  N := 0;
+  for SpeechSegment in AllSpeechSegment do
+    Inc(N, Length(SpeechSegment.Samples));
+
+  SetLength(AllSamples, N);
+
+  N := 0;
+  for SpeechSegment in AllSpeechSegment do
+    begin
+      for I := Low(SpeechSegment.Samples) to High(SpeechSegment.Samples) do
+        begin
+          AllSamples[N] := SpeechSegment.Samples[I];
+          Inc(N);
+        end;
+    end;
+
+  SherpaOnnxWriteWave('./lei-jun-test-no-silence-ten-vad.wav', AllSamples, SampleRate);
+  WriteLn('Saved to ./lei-jun-test-no-silence-ten-vad.wav');
+
+  FreeAndNil(Vad);
+end.

+#!/usr/bin/env bash
+
+set -ex
+
+SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
+SHERPA_ONNX_DIR=$(cd $SCRIPT_DIR/../.. && pwd)
+
+echo "SHERPA_ONNX_DIR: $SHERPA_ONNX_DIR"
+
+if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib  && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
+  mkdir -p ../../build
+  pushd ../../build
+  cmake \
+    -DCMAKE_INSTALL_PREFIX=./install \
+    -DSHERPA_ONNX_ENABLE_PYTHON=OFF \
+    -DSHERPA_ONNX_ENABLE_TESTS=OFF \
+    -DSHERPA_ONNX_ENABLE_CHECK=OFF \
+    -DBUILD_SHARED_LIBS=ON \
+    -DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
+    ..
+
+  cmake --build . --target install --config Release
+  popd
+fi
+
+if [[ ! -f ./ten-vad.onnx ]]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
+fi
+
+if [ ! -f ./lei-jun-test.wav ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
+fi
+
+fpc \
+  -dSHERPA_ONNX_USE_SHARED_LIBS \
+  -Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
+  -Fl$SHERPA_ONNX_DIR/build/install/lib \
+  ./remove_silence_ten_vad.pas
+
+export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
+export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
+
+./remove_silence_ten_vad

@@ -426,12 +426,24 @@ type
     class operator Initialize({$IFDEF FPC}var{$ELSE}out{$ENDIF} Dest: TSherpaOnnxSileroVadModelConfig);
   end;
 
+  TSherpaOnnxTenVadModelConfig = record
+    Model: AnsiString;
+    Threshold: Single;
+    MinSilenceDuration: Single;
+    MinSpeechDuration: Single;
+    WindowSize: Integer;
+    MaxSpeechDuration: Single;
+    function ToString: AnsiString;
+    class operator Initialize({$IFDEF FPC}var{$ELSE}out{$ENDIF} Dest: TSherpaOnnxTenVadModelConfig);
+  end;
+
   TSherpaOnnxVadModelConfig = record
     SileroVad: TSherpaOnnxSileroVadModelConfig;
     SampleRate: Integer;
     NumThreads: Integer;
     Provider: AnsiString;
     Debug: Boolean;
+    TenVad: TSherpaOnnxTenVadModelConfig;
     function ToString: AnsiString;
     class operator Initialize({$IFDEF FPC}var{$ELSE}out{$ENDIF} Dest: TSherpaOnnxVadModelConfig);
   end;
@@ -829,12 +841,23 @@ type
     WindowSize: cint32;
     MaxSpeechDuration: cfloat;
   end;
+
+  SherpaOnnxTenVadModelConfig = record
+    Model: PAnsiChar;
+    Threshold: cfloat;
+    MinSilenceDuration: cfloat;
+    MinSpeechDuration: cfloat;
+    WindowSize: cint32;
+    MaxSpeechDuration: cfloat;
+  end;
+
   SherpaOnnxVadModelConfig = record
     SileroVad: SherpaOnnxSileroVadModelConfig;
     SampleRate: cint32;
     NumThreads: cint32;
     Provider: PAnsiChar;
     Debug: cint32;
+    TenVad: SherpaOnnxTenVadModelConfig;
   end;
   PSherpaOnnxVadModelConfig = ^SherpaOnnxVadModelConfig;
 
@@ -1907,6 +1930,21 @@ begin
     ]);
 end;
 
+function TSherpaOnnxTenVadModelConfig.ToString: AnsiString;
+begin
+  Result := Format('TSherpaOnnxTenVadModelConfig(' +
+    'Model := %s, ' +
+    'Threshold := %.2f, ' +
+    'MinSilenceDuration := %.2f, ' +
+    'MinSpeechDuration := %.2f, ' +
+    'WindowSize := %d, ' +
+    'MaxSpeechDuration := %.2f' +
+    ')',
+    [Self.Model, Self.Threshold, Self.MinSilenceDuration,
+     Self.MinSpeechDuration, Self.WindowSize, Self.MaxSpeechDuration
+    ]);
+end;
+
 class operator TSherpaOnnxSileroVadModelConfig.Initialize({$IFDEF FPC}var{$ELSE}out{$ENDIF} Dest: TSherpaOnnxSileroVadModelConfig);
 begin
   Dest.Threshold := 0.5;
@@ -1916,6 +1954,15 @@ begin
   Dest.MaxSpeechDuration := 5.0;
 end;
 
+class operator TSherpaOnnxTenVadModelConfig.Initialize({$IFDEF FPC}var{$ELSE}out{$ENDIF} Dest: TSherpaOnnxTenVadModelConfig);
+begin
+  Dest.Threshold := 0.5;
+  Dest.MinSilenceDuration := 0.5;
+  Dest.MinSpeechDuration := 0.25;
+  Dest.WindowSize := 256;
+  Dest.MaxSpeechDuration := 5.0;
+end;
+
 function TSherpaOnnxVadModelConfig.ToString: AnsiString;
 begin
   Result := Format('TSherpaOnnxVadModelConfig(' +
@@ -1923,10 +1970,11 @@ begin
     'SampleRate := %d, ' +
     'NumThreads := %d, ' +
     'Provider := %s, ' +
-    'Debug := %s' +
+    'Debug := %s, ' +
+    'TenVad := %s' +
     ')',
     [Self.SileroVad.ToString, Self.SampleRate, Self.NumThreads, Self.Provider,
-     Self.Debug.ToString
+     Self.Debug.ToString, Self.TenVad.ToString
     ]);
 end;
 
@@ -2077,6 +2125,13 @@ begin
   C.SileroVad.WindowSize := Config.SileroVad.WindowSize;
   C.SileroVad.MaxSpeechDuration := Config.SileroVad.MaxSpeechDuration;
 
+  C.TenVad.Model := PAnsiChar(Config.TenVad.Model);
+  C.TenVad.Threshold := Config.TenVad.Threshold;
+  C.TenVad.MinSilenceDuration := Config.TenVad.MinSilenceDuration;
+  C.TenVad.MinSpeechDuration := Config.TenVad.MinSpeechDuration;
+  C.TenVad.WindowSize := Config.TenVad.WindowSize;
+  C.TenVad.MaxSpeechDuration := Config.TenVad.MaxSpeechDuration;
+
   C.SampleRate := Config.SampleRate;
   C.NumThreads := Config.NumThreads;
   C.Provider := PAnsiChar(Config.Provider);