Add non-streaming ASR APIs for node-addon-api (#868)

nodejs-addon-examples/README.md

@@ -39,7 +39,7 @@ npm install naudiodon2
 node ./test_vad_microphone.js
 ```
 
-## Streaming speech recognition with zipformer transducer
+## Streaming speech recognition with Zipformer transducer
 
 ```bash
 wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20.tar.bz2
@@ -54,7 +54,7 @@ npm install naudiodon2
 node ./test_asr_streaming_transducer_microphone.js
 ```
 
-## Streaming speech recognition with zipformer CTC
+## Streaming speech recognition with Zipformer CTC
 
 ```bash
 wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
@@ -72,3 +72,74 @@ npm install naudiodon2
 node ./test_asr_streaming_ctc_microphone.js
 node ./test_asr_streaming_ctc_hlg_microphone.js
 ```
+
+## Streaming speech recognition with Paraformer
+
+```bash
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2
+tar xvf sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2
+rm sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2
+
+node ./test_asr_streaming_paraformer.js
+
+# To run the test with a microphone, you need to install the package naudiodon2
+npm install naudiodon2
+
+node ./test_asr_streaming_paraformer_microphone.js
+```
+
+## Non-streaming speech recognition with Zipformer transducer
+
+```bash
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-zipformer-en-2023-04-01.tar.bz2
+tar xvf sherpa-onnx-zipformer-en-2023-04-01.tar.bz2
+rm sherpa-onnx-zipformer-en-2023-04-01.tar.bz2
+
+node ./test_asr_non_streaming_transducer.js
+
+# To run VAD + non-streaming ASR with transudcer using a microphone
+npm install naudiodon2
+node ./test_vad_asr_non_streaming_transducer_microphone.js
+```
+
+## Non-streaming speech recognition with Whisper
+
+```bash
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.en.tar.bz2
+tar xvf sherpa-onnx-whisper-tiny.en.tar.bz2
+rm sherpa-onnx-whisper-tiny.en.tar.bz2
+
+node ./test_asr_non_streaming_whisper.js
+
+# To run VAD + non-streaming ASR with Paraformer using a microphone
+npm install naudiodon2
+node ./test_vad_asr_non_streaming_whisper_microphone.js
+```
+
+## Non-streaming speech recognition with NeMo CTC models
+
+```bash
+wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k.tar.bz2
+tar xvf sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k.tar.bz2
+rm sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k.tar.bz2
+
+node ./test_asr_non_streaming_nemo_ctc.js
+
+# To run VAD + non-streaming ASR with Paraformer using a microphone
+npm install naudiodon2
+node ./test_vad_asr_non_streaming_nemo_ctc_microphone.js
+```
+
+## Non-streaming speech recognition with Paraformer
+
+```bash
+wget 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
+
+node ./test_asr_non_streaming_paraformer.js
+
+# To run VAD + non-streaming ASR with Paraformer using a microphone
+npm install naudiodon2
+node ./test_vad_asr_non_streaming_paraformer_microphone.js
+```

nodejs-addon-examples/test_asr_non_streaming_nemo_ctc.js

@@ -0,0 +1,50 @@
+// Copyright (c)  2024  Xiaomi Corporation
+const sherpa_onnx = require('sherpa-onnx-node');
+const performance = require('perf_hooks').performance;
+
+
+// Please download test files from
+// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+const config = {
+  'featConfig': {
+    'sampleRate': 16000,
+    'featureDim': 80,
+  },
+  'modelConfig': {
+    'nemoCtc': {
+      'model':
+          './sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k/model.onnx',
+    },
+    'tokens':
+        './sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k/tokens.txt',
+    'numThreads': 2,
+    'provider': 'cpu',
+    'debug': 1,
+  }
+};
+
+const waveFilename =
+    './sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k/test_wavs/de-german.wav';
+
+const recognizer = new sherpa_onnx.OfflineRecognizer(config);
+console.log('Started')
+let start = performance.now();
+const stream = recognizer.createStream();
+const wave = sherpa_onnx.readWave(waveFilename);
+stream.acceptWaveform({sampleRate: wave.sampleRate, samples: wave.samples});
+
+recognizer.decode(stream);
+result = recognizer.getResult(stream)
+let stop = performance.now();
+console.log('Done')
+
+const elapsed_seconds = (stop - start) / 1000;
+const duration = wave.samples.length / wave.sampleRate;
+const real_time_factor = elapsed_seconds / duration;
+console.log('Wave duration', duration.toFixed(3), 'secodns')
+console.log('Elapsed', elapsed_seconds.toFixed(3), 'secodns')
+console.log(
+    `RTF = ${elapsed_seconds.toFixed(3)}/${duration.toFixed(3)} =`,
+    real_time_factor.toFixed(3))
+console.log(waveFilename)
+console.log('result\n', result)

nodejs-addon-examples/test_asr_non_streaming_paraformer.js

@@ -0,0 +1,48 @@
+// Copyright (c)  2024  Xiaomi Corporation
+const sherpa_onnx = require('sherpa-onnx-node');
+const performance = require('perf_hooks').performance;
+
+
+// Please download test files from
+// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+const config = {
+  'featConfig': {
+    'sampleRate': 16000,
+    'featureDim': 80,
+  },
+  'modelConfig': {
+    'paraformer': {
+      'model': './sherpa-onnx-paraformer-zh-2023-03-28/model.int8.onnx',
+    },
+    'tokens': './sherpa-onnx-paraformer-zh-2023-03-28/tokens.txt',
+    'numThreads': 2,
+    'provider': 'cpu',
+    'debug': 1,
+  }
+};
+
+const waveFilename =
+    './sherpa-onnx-paraformer-zh-2023-03-28/test_wavs/5-henan.wav';
+
+const recognizer = new sherpa_onnx.OfflineRecognizer(config);
+console.log('Started')
+let start = performance.now();
+const stream = recognizer.createStream();
+const wave = sherpa_onnx.readWave(waveFilename);
+stream.acceptWaveform({sampleRate: wave.sampleRate, samples: wave.samples});
+
+recognizer.decode(stream);
+result = recognizer.getResult(stream)
+let stop = performance.now();
+console.log('Done')
+
+const elapsed_seconds = (stop - start) / 1000;
+const duration = wave.samples.length / wave.sampleRate;
+const real_time_factor = elapsed_seconds / duration;
+console.log('Wave duration', duration.toFixed(3), 'secodns')
+console.log('Elapsed', elapsed_seconds.toFixed(3), 'secodns')
+console.log(
+    `RTF = ${elapsed_seconds.toFixed(3)}/${duration.toFixed(3)} =`,
+    real_time_factor.toFixed(3))
+console.log(waveFilename)
+console.log('result\n', result)

nodejs-addon-examples/test_asr_non_streaming_transducer.js

@@ -0,0 +1,52 @@
+// Copyright (c)  2024  Xiaomi Corporation
+const sherpa_onnx = require('sherpa-onnx-node');
+const performance = require('perf_hooks').performance;
+
+
+// Please download test files from
+// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+const config = {
+  'featConfig': {
+    'sampleRate': 16000,
+    'featureDim': 80,
+  },
+  'modelConfig': {
+    'transducer': {
+      'encoder':
+          './sherpa-onnx-zipformer-en-2023-04-01/encoder-epoch-99-avg-1.int8.onnx',
+      'decoder':
+          './sherpa-onnx-zipformer-en-2023-04-01/decoder-epoch-99-avg-1.onnx',
+      'joiner':
+          './sherpa-onnx-zipformer-en-2023-04-01/joiner-epoch-99-avg-1.int8.onnx',
+    },
+    'tokens': './sherpa-onnx-zipformer-en-2023-04-01/tokens.txt',
+    'numThreads': 2,
+    'provider': 'cpu',
+    'debug': 1,
+  }
+};
+
+const waveFilename = './sherpa-onnx-zipformer-en-2023-04-01/test_wavs/1.wav';
+
+const recognizer = new sherpa_onnx.OfflineRecognizer(config);
+console.log('Started')
+let start = performance.now();
+const stream = recognizer.createStream();
+const wave = sherpa_onnx.readWave(waveFilename);
+stream.acceptWaveform({sampleRate: wave.sampleRate, samples: wave.samples});
+
+recognizer.decode(stream);
+result = recognizer.getResult(stream)
+let stop = performance.now();
+console.log('Done')
+
+const elapsed_seconds = (stop - start) / 1000;
+const duration = wave.samples.length / wave.sampleRate;
+const real_time_factor = elapsed_seconds / duration;
+console.log('Wave duration', duration.toFixed(3), 'secodns')
+console.log('Elapsed', elapsed_seconds.toFixed(3), 'secodns')
+console.log(
+    `RTF = ${elapsed_seconds.toFixed(3)}/${duration.toFixed(3)} =`,
+    real_time_factor.toFixed(3))
+console.log(waveFilename)
+console.log('result\n', result)

nodejs-addon-examples/test_asr_non_streaming_whisper.js

@@ -0,0 +1,48 @@
+// Copyright (c)  2024  Xiaomi Corporation
+const sherpa_onnx = require('sherpa-onnx-node');
+const performance = require('perf_hooks').performance;
+
+
+// Please download test files from
+// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+const config = {
+  'featConfig': {
+    'sampleRate': 16000,
+    'featureDim': 80,
+  },
+  'modelConfig': {
+    'whisper': {
+      'encoder': './sherpa-onnx-whisper-tiny.en/tiny.en-encoder.int8.onnx',
+      'decoder': './sherpa-onnx-whisper-tiny.en/tiny.en-decoder.int8.onnx',
+    },
+    'tokens': './sherpa-onnx-whisper-tiny.en/tiny.en-tokens.txt',
+    'numThreads': 2,
+    'provider': 'cpu',
+    'debug': 1,
+  }
+};
+
+const waveFilename = './sherpa-onnx-whisper-tiny.en/test_wavs/0.wav';
+
+const recognizer = new sherpa_onnx.OfflineRecognizer(config);
+console.log('Started')
+let start = performance.now();
+const stream = recognizer.createStream();
+const wave = sherpa_onnx.readWave(waveFilename);
+stream.acceptWaveform({sampleRate: wave.sampleRate, samples: wave.samples});
+
+recognizer.decode(stream);
+result = recognizer.getResult(stream)
+let stop = performance.now();
+console.log('Done')
+
+const elapsed_seconds = (stop - start) / 1000;
+const duration = wave.samples.length / wave.sampleRate;
+const real_time_factor = elapsed_seconds / duration;
+console.log('Wave duration', duration.toFixed(3), 'secodns')
+console.log('Elapsed', elapsed_seconds.toFixed(3), 'secodns')
+console.log(
+    `RTF = ${elapsed_seconds.toFixed(3)}/${duration.toFixed(3)} =`,
+    real_time_factor.toFixed(3))
+console.log(waveFilename)
+console.log('result\n', result)

nodejs-addon-examples/test_asr_streaming_paraformer.js

@@ -0,0 +1,56 @@
+// Copyright (c)  2024  Xiaomi Corporation
+const sherpa_onnx = require('sherpa-onnx-node');
+const performance = require('perf_hooks').performance;
+
+
+// Please download test files from
+// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+const config = {
+  'featConfig': {
+    'sampleRate': 16000,
+    'featureDim': 80,
+  },
+  'modelConfig': {
+    'paraformer': {
+      'encoder':
+          './sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx',
+      'decoder':
+          './sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx',
+    },
+    'tokens': './sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt',
+    'numThreads': 2,
+    'provider': 'cpu',
+    'debug': 1,
+  }
+};
+
+const waveFilename =
+    './sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/0.wav';
+
+const recognizer = new sherpa_onnx.OnlineRecognizer(config);
+console.log('Started')
+let start = performance.now();
+const stream = recognizer.createStream();
+const wave = sherpa_onnx.readWave(waveFilename);
+stream.acceptWaveform({sampleRate: wave.sampleRate, samples: wave.samples});
+
+const tailPadding = new Float32Array(wave.sampleRate * 0.4);
+stream.acceptWaveform({samples: tailPadding, sampleRate: wave.sampleRate});
+
+while (recognizer.isReady(stream)) {
+  recognizer.decode(stream);
+}
+result = recognizer.getResult(stream)
+let stop = performance.now();
+console.log('Done')
+
+const elapsed_seconds = (stop - start) / 1000;
+const duration = wave.samples.length / wave.sampleRate;
+const real_time_factor = elapsed_seconds / duration;
+console.log('Wave duration', duration.toFixed(3), 'secodns')
+console.log('Elapsed', elapsed_seconds.toFixed(3), 'secodns')
+console.log(
+    `RTF = ${elapsed_seconds.toFixed(3)}/${duration.toFixed(3)} =`,
+    real_time_factor.toFixed(3))
+console.log(waveFilename)
+console.log('result\n', result)

nodejs-addon-examples/test_asr_streaming_paraformer_microphone.js

@@ -0,0 +1,104 @@
+// Copyright (c)  2023-2024  Xiaomi Corporation (authors: Fangjun Kuang)
+//
+const portAudio = require('naudiodon2');
+// console.log(portAudio.getDevices());
+
+const sherpa_onnx = require('sherpa-onnx-node');
+
+function createOnlineRecognizer() {
+  const config = {
+    'featConfig': {
+      'sampleRate': 16000,
+      'featureDim': 80,
+    },
+    'modelConfig': {
+      'paraformer': {
+        'encoder':
+            './sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx',
+        'decoder':
+            './sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx',
+      },
+      'tokens': './sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt',
+      'numThreads': 2,
+      'provider': 'cpu',
+      'debug': 1,
+    },
+    'decodingMethod': 'greedy_search',
+    'maxActivePaths': 4,
+    'enableEndpoint': true,
+    'rule1MinTrailingSilence': 2.4,
+    'rule2MinTrailingSilence': 1.2,
+    'rule3MinUtteranceLength': 20
+  };
+
+  return new sherpa_onnx.OnlineRecognizer(config);
+}
+
+const recognizer = createOnlineRecognizer();
+const stream = recognizer.createStream();
+
+let lastText = '';
+let segmentIndex = 0;
+
+const ai = new portAudio.AudioIO({
+  inOptions: {
+    channelCount: 1,
+    closeOnError: true,  // Close the stream if an audio error is detected, if
+                         // set false then just log the error
+    deviceId: -1,  // Use -1 or omit the deviceId to select the default device
+    sampleFormat: portAudio.SampleFormatFloat32,
+    sampleRate: recognizer.config.featConfig.sampleRate
+  }
+});
+
+const display = new sherpa_onnx.Display(50);
+
+ai.on('data', data => {
+  const samples = new Float32Array(data.buffer);
+
+  stream.acceptWaveform(
+      {sampleRate: recognizer.config.featConfig.sampleRate, samples: samples});
+
+  while (recognizer.isReady(stream)) {
+    recognizer.decode(stream);
+  }
+
+  const isEndpoint = recognizer.isEndpoint(stream);
+  let text = recognizer.getResult(stream).text.toLowerCase();
+
+  if (isEndpoint) {
+    // for online paraformer models, we have to manually padding on endpoint
+    // so that the last word can be recognized
+    const tailPadding =
+        new Float32Array(recognizer.config.featConfig.sampleRate * 0.4);
+    stream.acceptWaveform({
+      samples: tailPadding,
+      sampleRate: recognizer.config.featConfig.sampleRate
+    });
+    while (recognizer.isReady(stream)) {
+      recognizer.decode(stream);
+    }
+    text = recognizer.getResult(stream).text.toLowerCase();
+  }
+
+  if (text.length > 0 && lastText != text) {
+    lastText = text;
+    display.print(segmentIndex, lastText);
+  }
+  if (isEndpoint) {
+    if (text.length > 0) {
+      lastText = text;
+      segmentIndex += 1;
+    }
+    recognizer.reset(stream)
+  }
+});
+
+ai.on('close', () => {
+  console.log('Free resources');
+  stream.free();
+  recognizer.free();
+});
+
+ai.start();
+console.log('Started! Please speak')

nodejs-addon-examples/test_vad_asr_non_streaming_nemo_ctc_microphone.js

@@ -0,0 +1,110 @@
+// Copyright (c)  2023-2024  Xiaomi Corporation (authors: Fangjun Kuang)
+//
+const portAudio = require('naudiodon2');
+// console.log(portAudio.getDevices());
+
+const sherpa_onnx = require('sherpa-onnx-node');
+
+function createRecognizer() {
+  // Please download test files from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+  const config = {
+    'featConfig': {
+      'sampleRate': 16000,
+      'featureDim': 80,
+    },
+    'modelConfig': {
+      'nemoCtc': {
+        'model':
+            './sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k/model.onnx',
+      },
+      'tokens':
+          './sherpa-onnx-nemo-fast-conformer-ctc-be-de-en-es-fr-hr-it-pl-ru-uk-20k/tokens.txt',
+      'numThreads': 2,
+      'provider': 'cpu',
+      'debug': 1,
+    }
+  };
+
+  return new sherpa_onnx.OfflineRecognizer(config);
+}
+
+function createVad() {
+  // please download silero_vad.onnx from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+  const config = {
+    sileroVad: {
+      model: './silero_vad.onnx',
+      threshold: 0.5,
+      minSpeechDuration: 0.25,
+      minSilenceDuration: 0.5,
+      windowSize: 512,
+    },
+    sampleRate: 16000,
+    debug: true,
+    numThreads: 1,
+  };
+
+  const bufferSizeInSeconds = 60;
+
+  return new sherpa_onnx.Vad(config, bufferSizeInSeconds);
+}
+
+const recognizer = createRecognizer();
+const vad = createVad();
+
+const bufferSizeInSeconds = 30;
+const buffer =
+    new sherpa_onnx.CircularBuffer(bufferSizeInSeconds * vad.config.sampleRate);
+
+const ai = new portAudio.AudioIO({
+  inOptions: {
+    channelCount: 1,
+    closeOnError: true,  // Close the stream if an audio error is detected, if
+                         // set false then just log the error
+    deviceId: -1,  // Use -1 or omit the deviceId to select the default device
+    sampleFormat: portAudio.SampleFormatFloat32,
+    sampleRate: vad.config.sampleRate
+  }
+});
+
+let printed = false;
+let index = 0;
+ai.on('data', data => {
+  const windowSize = vad.config.sileroVad.windowSize;
+  buffer.push(new Float32Array(data.buffer));
+  while (buffer.size() > windowSize) {
+    const samples = buffer.get(buffer.head(), windowSize);
+    buffer.pop(windowSize);
+    vad.acceptWaveform(samples);
+  }
+
+  while (!vad.isEmpty()) {
+    const segment = vad.front();
+    vad.pop();
+    const stream = recognizer.createStream();
+    stream.acceptWaveform({
+      samples: segment.samples,
+      sampleRate: recognizer.config.featConfig.sampleRate
+    });
+    recognizer.decode(stream);
+    const r = recognizer.getResult(stream);
+    if (r.text.length > 0) {
+      const text = r.text.toLowerCase().trim();
+      console.log(`${index}: ${text}`);
+
+      const filename = `${index}-${text}-${
+          new Date()
+              .toLocaleTimeString('en-US', {hour12: false})
+              .split(' ')[0]}.wav`;
+      sherpa_onnx.writeWave(
+          filename,
+          {samples: segment.samples, sampleRate: vad.config.sampleRate})
+
+      index += 1;
+    }
+  }
+});
+
+ai.start();
+console.log('Started! Please speak')

nodejs-addon-examples/test_vad_asr_non_streaming_paraformer_microphone.js

@@ -0,0 +1,108 @@
+// Copyright (c)  2023-2024  Xiaomi Corporation (authors: Fangjun Kuang)
+//
+const portAudio = require('naudiodon2');
+// console.log(portAudio.getDevices());
+
+const sherpa_onnx = require('sherpa-onnx-node');
+
+function createRecognizer() {
+  // Please download test files from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+  const config = {
+    'featConfig': {
+      'sampleRate': 16000,
+      'featureDim': 80,
+    },
+    'modelConfig': {
+      'paraformer': {
+        'model': './sherpa-onnx-paraformer-zh-2023-03-28/model.int8.onnx',
+      },
+      'tokens': './sherpa-onnx-paraformer-zh-2023-03-28/tokens.txt',
+      'numThreads': 2,
+      'provider': 'cpu',
+      'debug': 1,
+    }
+  };
+
+  return new sherpa_onnx.OfflineRecognizer(config);
+}
+
+function createVad() {
+  // please download silero_vad.onnx from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+  const config = {
+    sileroVad: {
+      model: './silero_vad.onnx',
+      threshold: 0.5,
+      minSpeechDuration: 0.25,
+      minSilenceDuration: 0.5,
+      windowSize: 512,
+    },
+    sampleRate: 16000,
+    debug: true,
+    numThreads: 1,
+  };
+
+  const bufferSizeInSeconds = 60;
+
+  return new sherpa_onnx.Vad(config, bufferSizeInSeconds);
+}
+
+const recognizer = createRecognizer();
+const vad = createVad();
+
+const bufferSizeInSeconds = 30;
+const buffer =
+    new sherpa_onnx.CircularBuffer(bufferSizeInSeconds * vad.config.sampleRate);
+
+const ai = new portAudio.AudioIO({
+  inOptions: {
+    channelCount: 1,
+    closeOnError: true,  // Close the stream if an audio error is detected, if
+                         // set false then just log the error
+    deviceId: -1,  // Use -1 or omit the deviceId to select the default device
+    sampleFormat: portAudio.SampleFormatFloat32,
+    sampleRate: vad.config.sampleRate
+  }
+});
+
+let printed = false;
+let index = 0;
+ai.on('data', data => {
+  const windowSize = vad.config.sileroVad.windowSize;
+  buffer.push(new Float32Array(data.buffer));
+  while (buffer.size() > windowSize) {
+    const samples = buffer.get(buffer.head(), windowSize);
+    buffer.pop(windowSize);
+    vad.acceptWaveform(samples);
+  }
+
+  while (!vad.isEmpty()) {
+    const segment = vad.front();
+    vad.pop();
+    const stream = recognizer.createStream();
+    stream.acceptWaveform({
+      samples: segment.samples,
+      sampleRate: recognizer.config.featConfig.sampleRate
+    });
+    recognizer.decode(stream);
+    const r = recognizer.getResult(stream);
+    if (r.text.length > 0) {
+      const text = r.text.toLowerCase().trim();
+      console.log(`${index}: ${text}`);
+
+      const filename = `${index}-${text}-${
+          new Date()
+              .toLocaleTimeString('en-US', {hour12: false})
+              .split(' ')[0]}.wav`;
+      sherpa_onnx.writeWave(
+          filename,
+          {samples: segment.samples, sampleRate: vad.config.sampleRate})
+
+      index += 1;
+    }
+  }
+});
+
+ai.start();
+console.log('Started! Please speak')

nodejs-addon-examples/test_vad_asr_non_streaming_transducer_microphone.js

@@ -0,0 +1,113 @@
+// Copyright (c)  2023-2024  Xiaomi Corporation (authors: Fangjun Kuang)
+//
+const portAudio = require('naudiodon2');
+// console.log(portAudio.getDevices());
+
+const sherpa_onnx = require('sherpa-onnx-node');
+
+function createRecognizer() {
+  // Please download test files from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+  const config = {
+    'featConfig': {
+      'sampleRate': 16000,
+      'featureDim': 80,
+    },
+    'modelConfig': {
+      'transducer': {
+        'encoder':
+            './sherpa-onnx-zipformer-en-2023-04-01/encoder-epoch-99-avg-1.int8.onnx',
+        'decoder':
+            './sherpa-onnx-zipformer-en-2023-04-01/decoder-epoch-99-avg-1.onnx',
+        'joiner':
+            './sherpa-onnx-zipformer-en-2023-04-01/joiner-epoch-99-avg-1.int8.onnx',
+      },
+      'tokens': './sherpa-onnx-zipformer-en-2023-04-01/tokens.txt',
+      'numThreads': 2,
+      'provider': 'cpu',
+      'debug': 1,
+    }
+  };
+
+  return new sherpa_onnx.OfflineRecognizer(config);
+}
+
+function createVad() {
+  // please download silero_vad.onnx from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+  const config = {
+    sileroVad: {
+      model: './silero_vad.onnx',
+      threshold: 0.5,
+      minSpeechDuration: 0.25,
+      minSilenceDuration: 0.5,
+      windowSize: 512,
+    },
+    sampleRate: 16000,
+    debug: true,
+    numThreads: 1,
+  };
+
+  const bufferSizeInSeconds = 60;
+
+  return new sherpa_onnx.Vad(config, bufferSizeInSeconds);
+}
+
+const recognizer = createRecognizer();
+const vad = createVad();
+
+const bufferSizeInSeconds = 30;
+const buffer =
+    new sherpa_onnx.CircularBuffer(bufferSizeInSeconds * vad.config.sampleRate);
+
+const ai = new portAudio.AudioIO({
+  inOptions: {
+    channelCount: 1,
+    closeOnError: true,  // Close the stream if an audio error is detected, if
+                         // set false then just log the error
+    deviceId: -1,  // Use -1 or omit the deviceId to select the default device
+    sampleFormat: portAudio.SampleFormatFloat32,
+    sampleRate: vad.config.sampleRate
+  }
+});
+
+let printed = false;
+let index = 0;
+ai.on('data', data => {
+  const windowSize = vad.config.sileroVad.windowSize;
+  buffer.push(new Float32Array(data.buffer));
+  while (buffer.size() > windowSize) {
+    const samples = buffer.get(buffer.head(), windowSize);
+    buffer.pop(windowSize);
+    vad.acceptWaveform(samples);
+  }
+
+  while (!vad.isEmpty()) {
+    const segment = vad.front();
+    vad.pop();
+    const stream = recognizer.createStream();
+    stream.acceptWaveform({
+      samples: segment.samples,
+      sampleRate: recognizer.config.featConfig.sampleRate
+    });
+    recognizer.decode(stream);
+    const r = recognizer.getResult(stream);
+    if (r.text.length > 0) {
+      const text = r.text.toLowerCase().trim();
+      console.log(`${index}: ${text}`);
+
+      const filename = `${index}-${text}-${
+          new Date()
+              .toLocaleTimeString('en-US', {hour12: false})
+              .split(' ')[0]}.wav`;
+      sherpa_onnx.writeWave(
+          filename,
+          {samples: segment.samples, sampleRate: vad.config.sampleRate})
+
+      index += 1;
+    }
+  }
+});
+
+ai.start();
+console.log('Started! Please speak')

nodejs-addon-examples/test_vad_asr_non_streaming_whisper_microphone.js

@@ -0,0 +1,109 @@
+// Copyright (c)  2023-2024  Xiaomi Corporation (authors: Fangjun Kuang)
+//
+const portAudio = require('naudiodon2');
+// console.log(portAudio.getDevices());
+
+const sherpa_onnx = require('sherpa-onnx-node');
+
+function createRecognizer() {
+  // Please download test files from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
+  const config = {
+    'featConfig': {
+      'sampleRate': 16000,
+      'featureDim': 80,
+    },
+    'modelConfig': {
+      'whisper': {
+        'encoder': './sherpa-onnx-whisper-tiny.en/tiny.en-encoder.int8.onnx',
+        'decoder': './sherpa-onnx-whisper-tiny.en/tiny.en-decoder.int8.onnx',
+      },
+      'tokens': './sherpa-onnx-whisper-tiny.en/tiny.en-tokens.txt',
+      'numThreads': 2,
+      'provider': 'cpu',
+      'debug': 1,
+    }
+  };
+
+  return new sherpa_onnx.OfflineRecognizer(config);
+}
+
+function createVad() {
+  // please download silero_vad.onnx from
+  // https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
+  const config = {
+    sileroVad: {
+      model: './silero_vad.onnx',
+      threshold: 0.5,
+      minSpeechDuration: 0.25,
+      minSilenceDuration: 0.5,
+      windowSize: 512,
+    },
+    sampleRate: 16000,
+    debug: true,
+    numThreads: 1,
+  };
+
+  const bufferSizeInSeconds = 60;
+
+  return new sherpa_onnx.Vad(config, bufferSizeInSeconds);
+}
+
+const recognizer = createRecognizer();
+const vad = createVad();
+
+const bufferSizeInSeconds = 30;
+const buffer =
+    new sherpa_onnx.CircularBuffer(bufferSizeInSeconds * vad.config.sampleRate);
+
+const ai = new portAudio.AudioIO({
+  inOptions: {
+    channelCount: 1,
+    closeOnError: true,  // Close the stream if an audio error is detected, if
+                         // set false then just log the error
+    deviceId: -1,  // Use -1 or omit the deviceId to select the default device
+    sampleFormat: portAudio.SampleFormatFloat32,
+    sampleRate: vad.config.sampleRate
+  }
+});
+
+let printed = false;
+let index = 0;
+ai.on('data', data => {
+  const windowSize = vad.config.sileroVad.windowSize;
+  buffer.push(new Float32Array(data.buffer));
+  while (buffer.size() > windowSize) {
+    const samples = buffer.get(buffer.head(), windowSize);
+    buffer.pop(windowSize);
+    vad.acceptWaveform(samples);
+  }
+
+  while (!vad.isEmpty()) {
+    const segment = vad.front();
+    vad.pop();
+    const stream = recognizer.createStream();
+    stream.acceptWaveform({
+      samples: segment.samples,
+      sampleRate: recognizer.config.featConfig.sampleRate
+    });
+    recognizer.decode(stream);
+    const r = recognizer.getResult(stream);
+    if (r.text.length > 0) {
+      const text = r.text.toLowerCase().trim();
+      console.log(`${index}: ${text}`);
+
+      const filename = `${index}-${text}-${
+          new Date()
+              .toLocaleTimeString('en-US', {hour12: false})
+              .split(' ')[0]}.wav`;
+      sherpa_onnx.writeWave(
+          filename,
+          {samples: segment.samples, sampleRate: vad.config.sampleRate})
+
+      index += 1;
+    }
+  }
+});
+
+ai.start();
+console.log('Started! Please speak')