Non-Streaming ASR API

Non-streaming (offline) speech recognition API reference for sherpa-onnx-node.

Source file

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

API

OfflineRecognizer

Non-streaming speech recognizer. Processes a complete audio file at once.

Constructor

const recognizer = new sherpa_onnx.OfflineRecognizer(config);
param config

Recognizer configuration object (OfflineRecognizerConfig).

The config object supports:

  • featConfig (object, optional) — { sampleRate: number, featureDim: number }

  • modelConfig (object, optional) — Model configuration with one of:

    • transducer — { encoder: string, decoder: string, joiner: string }

    • paraformer — { encoder: string, decoder: string }

    • zipformerCtc — { model: string }

    • nemoCtc — { model: string }

    • senseVoice — { model: string }

    • whisper — { encoder: string, decoder: string, language: string, task: string }

    • moonshine — { encoder: string, decoder: string, uncachedDecoder: string }

    • fireRedAsr — { encoder: string, decoder: string }

    • cohereTranscribe — { model: string }

    Plus common fields: tokens, numThreads, debug, provider.

Static Methods

OfflineRecognizer.createAsync(config)

Create a recognizer asynchronously (non-blocking).

param config

Recognizer configuration (OfflineRecognizerConfig).

returns

A Promise<OfflineRecognizer>.

Methods

recognizer.createStream()
returns

A new OfflineStream.

recognizer.decode(stream)

Decode the stream synchronously.

param stream

An OfflineStream.

recognizer.decodeAsync(stream)

Decode the stream asynchronously (non-blocking).

param stream

An OfflineStream.

returns

A Promise<OfflineRecognizerResult>.

recognizer.getResult(stream)

Get the recognition result.

param stream

An OfflineStream.

returns

An OfflineRecognizerResult object with:

  • text (string) — Recognized text.

  • tokens (string[]) — Token strings.

  • timestamps (number[]) — Per-token timestamps in seconds.

  • durations (number[]) — Per-token durations in seconds.

  • lang (string) — Detected language (SenseVoice).

  • emotion (string) — Detected emotion (SenseVoice).

  • event (string) — Detected event (SenseVoice).

recognizer.setConfig(config)

Update the recognizer configuration at runtime.

param config

New configuration (OfflineRecognizerConfig).

Properties

  • recognizer.config — The configuration object.

OfflineStream

A non-streaming recognition stream.

Methods

stream.acceptWaveform(obj)

Feed audio to the stream.

param obj

{ samples: Float32Array, sampleRate: number }.

stream.setOption(key, value)

Set a string option on the stream.

param key

Option name (string).

param value

Option value (string).

Example

const sherpa_onnx = require('sherpa-onnx-node');

const recognizer = new sherpa_onnx.OfflineRecognizer({ /* config */ });
const stream = recognizer.createStream();

const wave = sherpa_onnx.readWave('./audio.wav');
stream.acceptWaveform({ samples: wave.samples, sampleRate: wave.sampleRate });

recognizer.decode(stream);
const result = recognizer.getResult(stream);
console.log(result.text);

Notes

  • The entire audio must be available before calling decode().

  • Use createAsync() for non-blocking construction in async contexts.

  • Use decodeAsync() for non-blocking decoding.