VAD with paraformer example

.github/scripts/test-dart.sh

@@ -6,6 +6,10 @@ cd dart-api-examples
 
 pushd non-streaming-asr
 
+echo '----------VAD with paraformer----------'
+./run-vad-with-paraformer.sh
+rm -rf sherpa-onnx-*
+
 echo '----------NeMo transducer----------'
 ./run-nemo-transducer.sh
 rm -rf sherpa-onnx-*

dart-api-examples/non-streaming-asr/README.md

@@ -1,2 +1,14 @@
-A sample command-line application with an entrypoint in `bin/`, library code
-in `lib/`, and example unit test in `test/`.
+# Introduction
+
+This folder contains examples for non-streaming ASR with Dart API.
+
+| File | Description|
+|------|------------|
+|[./bin/nemo-ctc.dart](./bin/nemo-ctc.dart)| Use a NeMo Ctc model for speech recognition. See [./run-nemo-ctc.sh](./run-nemo-ctc.sh)|
+|[./bin/nemo-transducer.dart](./bin/nemo-transducer.dart)| Use a NeMo transducer model for speech recognition. See [./run-nemo-transducer.sh](./run-nemo-transducer.sh)|
+|[./bin/paraformer.dart](./bin/paraformer.dart)|Use a paraformer model for speech recognition. See [./run-paraformer.sh](./run-paraformer.sh)|
+|[./bin/telespeech-ctc.dart](./bin/telespeech-ctc.dart)| Use models from [Tele-AI/TeleSpeech-ASR](https://github.com/Tele-AI/TeleSpeech-ASR) for speech recognition. See [./run-telespeech-ctc.sh](./run-telespeech-ctc.sh)|
+|[./bin/whisper.dart](./bin/whisper.dart)| Use whisper for speech recognition. See [./run-whisper.sh](./run-whisper.sh)|
+|[./bin/zipformer-transducer.dart](./bin/zipformer-transducer.dart)| Use a zipformer transducer for speech recognition. See [./run-zipformer-transducer.sh](./run-zipformer-transducer.sh)|
+|[./bin/vad-with-paraformer.dart](./bin/vad-with-paraformer.dart)| Use a [silero-vad](https://github.com/snakers4/silero-vad) with paraformer for speech recognition. See [./run-vad-with-paraformer.sh](./run-vad-with-paraformer.sh)|
+

dart-api-examples/non-streaming-asr/bin/vad-with-paraformer.dart

@@ -0,0 +1,97 @@
+import 'dart:io';
+import 'dart:typed_data';
+
+import 'package:args/args.dart';
+import 'package:sherpa_onnx/sherpa_onnx.dart' as sherpa_onnx;
+
+import './init.dart';
+
+void main(List<String> arguments) async {
+  await initSherpaOnnx();
+
+  final parser = ArgParser()
+    ..addOption('silero-vad', help: 'Path to silero_vad.onnx')
+    ..addOption('model', help: 'Path to the paraformer model')
+    ..addOption('tokens', help: 'Path to tokens.txt')
+    ..addOption('input-wav', help: 'Path to input.wav to transcribe');
+
+  final res = parser.parse(arguments);
+  if (res['silero-vad'] == null ||
+      res['model'] == null ||
+      res['tokens'] == null ||
+      res['input-wav'] == null) {
+    print(parser.usage);
+    exit(1);
+  }
+
+  final sileroVad = res['silero-vad'] as String;
+  final model = res['model'] as String;
+  final tokens = res['tokens'] as String;
+  final inputWav = res['input-wav'] as String;
+
+  final paraformer = sherpa_onnx.OfflineParaformerModelConfig(
+    model: model,
+  );
+
+  final modelConfig = sherpa_onnx.OfflineModelConfig(
+    paraformer: paraformer,
+    tokens: tokens,
+    debug: true,
+    numThreads: 1,
+    modelType: 'paraformer',
+  );
+  final config = sherpa_onnx.OfflineRecognizerConfig(model: modelConfig);
+  final recognizer = sherpa_onnx.OfflineRecognizer(config);
+
+  final sileroVadConfig = sherpa_onnx.SileroVadModelConfig(
+    model: sileroVad,
+    minSilenceDuration: 0.25,
+    minSpeechDuration: 0.5,
+  );
+
+  final vadConfig = sherpa_onnx.VadModelConfig(
+    sileroVad: sileroVadConfig,
+    numThreads: 1,
+    debug: true,
+  );
+
+  final vad = sherpa_onnx.VoiceActivityDetector(
+      config: vadConfig, bufferSizeInSeconds: 10);
+
+  final waveData = sherpa_onnx.readWave(inputWav);
+
+  int numSamples = waveData.samples.length;
+  int numIter = numSamples ~/ vadConfig.sileroVad.windowSize;
+
+  for (int i = 0; i != numIter; ++i) {
+    int start = i * vadConfig.sileroVad.windowSize;
+    vad.acceptWaveform(Float32List.sublistView(
+        waveData.samples, start, start + vadConfig.sileroVad.windowSize));
+
+    if (vad.isDetected()) {
+      while (!vad.isEmpty()) {
+        final stream = recognizer.createStream();
+        final segment = vad.front();
+        stream.acceptWaveform(
+            samples: segment.samples, sampleRate: waveData.sampleRate);
+        recognizer.decode(stream);
+
+        final result = recognizer.getResult(stream);
+
+        final startTime = segment.start * 1.0 / waveData.sampleRate;
+        final duration = segment.samples.length * 1.0 / waveData.sampleRate;
+        final stopTime = startTime + duration;
+        if (result.text != '') {
+          print(
+              '${startTime.toStringAsPrecision(4)} -- ${stopTime.toStringAsPrecision(4)}: ${result.text}');
+        }
+
+        stream.free();
+        vad.pop();
+      }
+    }
+  }
+
+  vad.free();
+  recognizer.free();
+}

dart-api-examples/non-streaming-asr/run-vad-with-paraformer.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+
+set -ex
+
+dart pub get
+
+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
+
+if [ ! -f ./sherpa-onnx-paraformer-zh-2023-03-28/tokens.txt ]; then
+  curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+
+  tar xvf sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+  rm sherpa-onnx-paraformer-zh-2023-03-28.tar.bz2
+fi
+
+dart run \
+  ./bin/vad-with-paraformer.dart \
+  --silero-vad ./silero_vad.onnx \
+  --model ./sherpa-onnx-paraformer-zh-2023-03-28/model.int8.onnx \
+  --tokens ./sherpa-onnx-paraformer-zh-2023-03-28/tokens.txt \
+  --input-wav ./lei-jun-test.wav