Inverse text normalization API of streaming ASR for various programming languages (#1022)

nodejs-addon-examples/package.json

@@ -1,5 +1,5 @@
 {
   "dependencies": {
-    "sherpa-onnx-node": "^1.9.30"
+    "sherpa-onnx-node": "^1.10.0"
   }
 }

nodejs-addon-examples/test_asr_streaming_transducer_itn.js

@@ -0,0 +1,59 @@
+// Copyright (c)  2024  Xiaomi Corporation
+const sherpa_onnx = require('sherpa-onnx-node');
+
+// 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-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,
+  },
+  // https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn_zh_number.fst
+  ruleFsts: './itn_zh_number.fst',
+};
+
+// https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn-zh-number.wav
+const waveFilename = './itn-zh-number.wav';
+
+const recognizer = new sherpa_onnx.OnlineRecognizer(config);
+console.log('Started')
+let start = Date.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 = Date.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_transducer_microphone_itn.js

@@ -0,0 +1,88 @@
+// 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,
+    },
+    'decodingMethod': 'greedy_search',
+    'maxActivePaths': 4,
+    'enableEndpoint': true,
+    'rule1MinTrailingSilence': 2.4,
+    'rule2MinTrailingSilence': 1.2,
+    'rule3MinUtteranceLength': 20,
+    // https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/itn_zh_number.fst
+    ruleFsts: './itn_zh_number.fst',
+  };
+
+  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);
+  const 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.start();
+console.log('Started! Please speak')