JavaScript (Node.js)
This page shows how to use X-ASR models with the JavaScript (Node.js) API of sherpa-onnx. Two demos are given: one with a non-streaming model and one with a streaming model.
On this page
Requirements
Node.js 18 or above
Non-streaming X-ASR
Step 1. Create the project directory
mkdir x-asr-demo
cd x-asr-demo
Step 2. Install the package
npm install sherpa-onnx-node
Step 3. Download the model
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03.tar.bz2
tar xf sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03.tar.bz2
rm sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03.tar.bz2
Step 4. Save the code
Save the following as x-asr-offline.js:
// This file shows how to use a non-streaming X-ASR model with sherpa-onnx
// in JavaScript (Node.js).
//
// Before you run it, please
//
// npm install sherpa-onnx-node
//
// and download the model:
//
// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03.tar.bz2
// tar xf sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03.tar.bz2
const sherpa_onnx = require('sherpa-onnx-node');
const modelDir =
'./sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03';
// Step 1: Create the recognizer from the X-ASR model files.
const config = {
featConfig: {sampleRate: 16000, featureDim: 80},
modelConfig: {
transducer: {
encoder: `${modelDir}/encoder-epoch-99-avg-1.int8.onnx`,
decoder: `${modelDir}/decoder-epoch-99-avg-1.onnx`,
joiner: `${modelDir}/joiner-epoch-99-avg-1.int8.onnx`,
},
tokens: `${modelDir}/tokens.txt`,
numThreads: 2,
provider: 'cpu',
debug: 0,
},
};
const recognizer = new sherpa_onnx.OfflineRecognizer(config);
// Step 2: Read the wave file you want to transcribe.
const filename = `${modelDir}/test_wavs/0.wav`;
const wave = sherpa_onnx.readWave(filename);
// Step 3: Create a stream, send the whole audio to it, and decode.
const start = Date.now();
const stream = recognizer.createStream();
stream.acceptWaveform({sampleRate: wave.sampleRate, samples: wave.samples});
recognizer.decode(stream);
// Step 4: Print the result. X-ASR punct models also add punctuations.
console.log(filename);
const elapsed = (Date.now() - start) / 1000;
console.log(recognizer.getResult(stream).text);
const audioDuration = wave.samples.length / wave.sampleRate;
console.log(`audio duration: ${audioDuration.toFixed(3)} s`);
console.log(`elapsed seconds: ${elapsed.toFixed(3)} s`);
console.log(
`RTF = ${elapsed.toFixed(3)}/${audioDuration.toFixed(3)} = ${(elapsed / audioDuration).toFixed(3)}`);
Step 5. Check the project layout
Your project directory should now look like this:
$ tree -I node_modules .
.
├── package-lock.json
├── package.json
├── sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03
│ ├── bpe.model
│ ├── decoder-epoch-99-avg-1.onnx
│ ├── encoder-epoch-99-avg-1.int8.onnx
│ ├── joiner-epoch-99-avg-1.int8.onnx
│ ├── README.md
│ ├── test_onnx.py
│ ├── test_wavs
│ │ ├── 0.wav
│ │ ├── 1.wav
│ │ ├── 2.wav
│ │ └── 3.wav
│ └── tokens.txt
└── x-asr-offline.js
3 directories, 14 files
$ ls -lh
total 24
drwxr-xr-x@ 5 fangjun wheel 160B 19 Nov 12:23 node_modules
-rw-r--r--@ 1 fangjun wheel 3.8K 19 Nov 12:23 package-lock.json
-rw-r--r--@ 1 fangjun wheel 61B 19 Nov 12:23 package.json
drwxr-xr-x@ 10 fangjun wheel 320B 29 Jun 16:31 sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03
-rw-r--r--@ 1 fangjun wheel 1.9K 19 Nov 12:23 x-asr-offline.js
Step 6. Build and run
node ./x-asr-offline.js
Output
./sherpa-onnx-x-asr-zipformer-transducer-zh-en-punct-int8-2026-06-03/test_wavs/0.wav
昨天是 Monday, today is 礼拜二, the day after tomorrow 是星期三。
audio duration: 10.053 s
elapsed seconds: 0.118 s
RTF = 0.118/10.053 = 0.012
Note
The timing above was measured on a MacBook Pro (Apple M4) using 2 threads. Please increase the number of threads for a smaller RTF on your machine.
Streaming X-ASR
Step 1. Create the project directory
mkdir x-asr-demo
cd x-asr-demo
Step 2. Install the package
npm install sherpa-onnx-node
Step 3. Download the model
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05.tar.bz2
tar xf sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05.tar.bz2
rm sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05.tar.bz2
Step 4. Save the code
Save the following as x-asr-streaming.js:
// This file shows how to use a streaming X-ASR model with sherpa-onnx
// in JavaScript (Node.js).
//
// Before you run it, please
//
// npm install sherpa-onnx-node
//
// and download the model:
//
// wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05.tar.bz2
// tar xf sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05.tar.bz2
const sherpa_onnx = require('sherpa-onnx-node');
const modelDir =
'./sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05';
// Step 1: Create the recognizer from the X-ASR model files.
const config = {
featConfig: {sampleRate: 16000, featureDim: 80},
modelConfig: {
transducer: {
encoder: `${modelDir}/encoder.int8.onnx`,
decoder: `${modelDir}/decoder.onnx`,
joiner: `${modelDir}/joiner.int8.onnx`,
},
tokens: `${modelDir}/tokens.txt`,
numThreads: 2,
provider: 'cpu',
debug: 0,
},
};
const recognizer = new sherpa_onnx.OnlineRecognizer(config);
// Step 2: Read the wave file you want to transcribe.
const filename = `${modelDir}/test_wavs/0.wav`;
const wave = sherpa_onnx.readWave(filename);
// Step 3: Create a stream and feed the audio in small chunks, just like
// a microphone would do in a real-time application.
const start = Date.now();
const stream = recognizer.createStream();
// 0.2 second for 16 kHz audio
const chunk = 3200;
let last = '';
for (let i = 0; i < wave.samples.length; i += chunk) {
stream.acceptWaveform({
sampleRate: wave.sampleRate,
samples: wave.samples.subarray(i, Math.min(i + chunk, wave.samples.length)),
});
while (recognizer.isReady(stream)) {
recognizer.decode(stream);
}
// Print the latest partial result when it changes.
const text = recognizer.getResult(stream).text;
if (text.length > 0 && text !== last) {
last = text;
console.log(text);
}
}
// Send some tail padding so the last words are flushed out.
const tailPadding = new Float32Array(wave.sampleRate * 0.5);
stream.acceptWaveform({samples: tailPadding, sampleRate: wave.sampleRate});
while (recognizer.isReady(stream)) {
recognizer.decode(stream);
}
// Step 4: Print the final result.
const elapsed = (Date.now() - start) / 1000;
console.log('Final result:', recognizer.getResult(stream).text);
const audioDuration = wave.samples.length / wave.sampleRate;
console.log(`audio duration: ${audioDuration.toFixed(3)} s`);
console.log(`elapsed seconds: ${elapsed.toFixed(3)} s`);
console.log(
`RTF = ${elapsed.toFixed(3)}/${audioDuration.toFixed(3)} = ${(elapsed / audioDuration).toFixed(3)}`);
Step 5. Check the project layout
Your project directory should now look like this:
$ tree -I node_modules .
.
├── package-lock.json
├── package.json
├── sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05
│ ├── bpe.model
│ ├── decoder.onnx
│ ├── encoder.int8.onnx
│ ├── joiner.int8.onnx
│ ├── README.md
│ ├── test_onnx.py
│ ├── test_wavs
│ │ ├── 0.wav
│ │ ├── 1.wav
│ │ ├── 2.wav
│ │ └── 3.wav
│ └── tokens.txt
└── x-asr-streaming.js
3 directories, 14 files
$ ls -lh
total 24
drwxr-xr-x@ 5 fangjun wheel 160B 19 Nov 12:23 node_modules
-rw-r--r--@ 1 fangjun wheel 3.8K 19 Nov 12:23 package-lock.json
-rw-r--r--@ 1 fangjun wheel 61B 19 Nov 12:23 package.json
drwxr-xr-x@ 10 fangjun wheel 320B 29 Jun 16:30 sherpa-onnx-x-asr-480ms-streaming-zipformer-transducer-zh-en-punct-int8-2026-06-05
-rw-r--r--@ 1 fangjun wheel 2.6K 19 Nov 12:23 x-asr-streaming.js
Step 6. Build and run
node ./x-asr-streaming.js
Output
昨
昨天是
昨天是 Mon
昨天是 Monday
昨天是 Monday,
昨天是 Monday, today
昨天是 Monday, today is 礼
昨天是 Monday, today is 礼拜二
昨天是 Monday, today is 礼拜二,
昨天是 Monday, today is 礼拜二, the day
昨天是 Monday, today is 礼拜二, the day after
昨天是 Monday, today is 礼拜二, the day after tomorro
昨天是 Monday, today is 礼拜二, the day after tomorrow
昨天是 Monday, today is 礼拜二, the day after tomorrow 是
Final result: 昨天是 Monday, today is 礼拜二, the day after tomorrow 是星期
audio duration: 10.053 s
elapsed seconds: 0.302 s
RTF = 0.302/10.053 = 0.030
Note
The timing above was measured on a MacBook Pro (Apple M4) using 2 threads. Please increase the number of threads for a smaller RTF on your machine.