Add more streaming ASR methods for node-addon-api

nodejs-addon-examples/README.md

@@ -28,9 +28,13 @@ export LD_LIBRARY_PATH=$PWD/node_modules/sherpa-onnx-linux-arm64:$LD_LIBRARY_PAT
 ```
 
 ## 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
 tar xvf sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20.tar.bz2
 rm sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20.tar.bz2
 
+node ./test_asr_streaming_transducer.js
+
+node ./test_asr_streaming_transducer_microphone.js
 ```

nodejs-addon-examples/package.json

@@ -1,6 +1,7 @@
 {
   "dependencies": {
-    "sherpa-onnx-node": "*",
-    "perf_hooks": "*"
+    "naudiodon2": "^2.4.0",
+    "perf_hooks": "*",
+    "sherpa-onnx-node": "*"
   }
 }

nodejs-addon-examples/test_asr_streaming_transducer_microphone.js

@@ -0,0 +1,92 @@
+// 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': {
+      'transducer': {
+        'encoder':
+            './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/encoder-epoch-99-avg-1.onnx',
+        'decoder':
+            './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/decoder-epoch-99-avg-1.onnx',
+        'joiner':
+            './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/joiner-epoch-99-avg-1.onnx',
+      },
+      'tokens':
+          './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/tokens.txt',
+      'numThreads': 2,
+      'provider': 'cpu',
+      'debug': 1,
+      'modelType': 'zipformer',
+    },
+    '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(samples, recognizer.config.featConfig.sampleRate);
+
+  while (recognizer.isReady(stream)) {
+    recognizer.decode(stream);
+  }
+
+  const isEndpoint = recognizer.isEndpoint(stream);
+  const text = recognizer.getResult(stream).text;
+
+  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')

scripts/node-addon-api/lib/sherpa-onnx.js

@@ -4,4 +4,5 @@ const streaming_asr = require('./streaming-asr.js');
 module.exports = {
   OnlineRecognizer: streaming_asr.OnlineRecognizer,
   readWave: addon.readWave,
+  Display: streaming_asr.Display,
 }

scripts/node-addon-api/lib/streaming-asr.js

@@ -1,5 +1,15 @@
 const addon = require('./addon.js');
 
+class Display {
+  constructor(maxWordPerline) {
+    this.handle = addon.createDisplay(maxWordPerline);
+  }
+
+  print(idx, text) {
+    addon.print(this.handle, idx, text)
+  }
+}
+
 class OnlineStream {
   constructor(handle) {
     this.handle = handle;
@@ -10,11 +20,16 @@ class OnlineStream {
     addon.acceptWaveformOnline(
         this.handle, {samples: samples, sampleRate: sampleRate})
   }
+
+  inputFinished() {
+    addon.inputFinished(this.handle)
+  }
 }
 
 class OnlineRecognizer {
   constructor(config) {
     this.handle = addon.createOnlineRecognizer(config);
+    this.config = config
   }
 
   createStream() {
@@ -30,6 +45,14 @@ class OnlineRecognizer {
     addon.decodeOnlineStream(this.handle, stream.handle);
   }
 
+  isEndpoint(stream) {
+    return addon.isEndpoint(this.handle, stream.handle);
+  }
+
+  reset(stream) {
+    addon.reset(this.handle, stream.handle);
+  }
+
   getResult(stream) {
     const jsonStr =
         addon.getOnlineStreamResultAsJson(this.handle, stream.handle);
@@ -38,4 +61,7 @@ class OnlineRecognizer {
   }
 }
 
-module.exports = {OnlineRecognizer}
+module.exports = {
+  OnlineRecognizer,
+  Display
+}