Add Pascal API for VAD

.github/workflows/pascal.yaml

@@ -116,12 +116,32 @@ jobs:
             cp -v install/lib/*.dll ../pascal-api-examples/read-wav
             cp -v install/lib/*.dll ../pascal-api-examples/streaming-asr
             cp -v install/lib/*.dll ../pascal-api-examples/non-streaming-asr
+            cp -v install/lib/*.dll ../pascal-api-examples/vad
 
             cp -v ../sherpa-onnx/pascal-api/sherpa_onnx.pas ../pascal-api-examples/read-wav
             cp -v ../sherpa-onnx/pascal-api/sherpa_onnx.pas ../pascal-api-examples/streaming-asr
             cp -v ../sherpa-onnx/pascal-api/sherpa_onnx.pas ../pascal-api-examples/non-streaming-asr
+            cp -v ../sherpa-onnx/pascal-api/sherpa_onnx.pas ../pascal-api-examples/vad
           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 "---"
+
+          ./run-remove-silence.sh
+          echo "---"
+
+          ls -lh
+
+          popd
+
       - name:  Run Pascal test (Read wav test)
         shell: bash
         run: |

pascal-api-examples/README.md

@@ -8,3 +8,5 @@ APIs of [sherpa-onnx](https://github.com/k2-fsa/sherpa-onnx).
 |[read-wav](./read-wav)|It shows how to read a wave file.|
 |[streaming-asr](./streaming-asr)| It shows how to use streaming models for speech recognition.|
 |[non-streaming-asr](./non-streaming-asr)| It shows how to use non-streaming models for speech recognition.|
+|[vad](./vad)| It shows how to use the voice activity detection API.|
+|[vad-with-non-streaming-asr](./vad-with-non-streaming-asr)| It shows how to use the voice activity detection API with non-streaming models for speech recognition.|

pascal-api-examples/vad/.gitignore

@@ -1,2 +1,3 @@
 !run*.sh
 circular_buffer
+remove_silence

pascal-api-examples/vad/README.md

@@ -0,0 +1,11 @@
+# Introduction
+
+
+This directory contains examples for how to use the VAD (voice activity detection)
+APIs.
+
+|Directory| Description|
+|---------|------------|
+|[run-circular-buffer.sh](./run-circular-buffer.sh)|It shows how to use the circular buffer API.|
+|[run-remove-silence.sh](./run-remove-silence.sh)|It shows how to use the VAD API to remove silences from a wave file.|
+

pascal-api-examples/vad/circular_buffer.pas

@@ -1,5 +1,8 @@
 { Copyright (c)  2024  Xiaomi Corporation }
 program circular_buffer;
+{
+This file shows how to use the CircularBuffer API of sherpa-onnx
+}
 
 {$mode objfpc}
 {$ASSERTIONS ON}

pascal-api-examples/vad/remove_silence.pas

@@ -0,0 +1,100 @@
+{ Copyright (c)  2024  Xiaomi Corporation }
+{
+This file shows how to use the VAD API from sherpa-onnx
+to remove silences from a wave file.
+}
+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 := 512; {Please don't change it unless you know the details}
+  Initialize(Config);
+
+  Config.SileroVad.Model := './silero_vad.onnx';
+  Config.SileroVad.MinSpeechDuration := 0.25;
+  Config.SileroVad.MinSilenceDuration := 0.5;
+  Config.SileroVad.Threshold := 0.5;
+  Config.SileroVad.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;
+
+  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.wav', AllSamples, SampleRate);
+  WriteLn('Saved to ./lei-jun-test-no-silence.wav');
+
+end.

pascal-api-examples/vad/run-remove-silence.sh

@@ -0,0 +1,42 @@
+#!/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: $SHERPA_ONNX_DIR"
+
+if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib  && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
+  mkdir -p ../../build
+  pushd ../../build
+  cmake \
+    -DCMAKE_INSTALL_PREFIX=./install \
+    -DSHERPA_ONNX_ENABLE_PYTHON=OFF \
+    -DSHERPA_ONNX_ENABLE_TESTS=OFF \
+    -DSHERPA_ONNX_ENABLE_CHECK=OFF \
+    -DBUILD_SHARED_LIBS=ON \
+    -DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
+    ..
+
+  cmake --build . --target install --config Release
+  popd
+fi
+
+if [[ ! -f ./silero_vad.onnx ]]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+fi
+
+if [ ! -f ./lei-jun-test.wav ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
+fi
+
+fpc \
+  -Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
+  -Fl$SHERPA_ONNX_DIR/build/install/lib \
+  ./remove_silence.pas
+
+export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
+export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
+
+./remove_silence