Audio Tagging API

Audio tagging API reference for sherpa-onnx-node.

Source file

scripts/node-addon-api/lib/audio-tagg.js

API

AudioTagging

Classifies audio into predefined event categories.

Constructor

const tagger = new sherpa_onnx.AudioTagging(config);

param config

Configuration object with:

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

  • ced (string) — Path to the CED ONNX model.

  • zipformer — { model: string } path to the ZipFormer ONNX model.

  Plus common fields: numThreads, debug, provider.

• labels (string, optional) — Path to the labels file.

• topK (number, optional) — Number of top results to return.

Methods

tagger.createStream()

returns

A new OfflineStream for feeding audio.

tagger.compute(stream, topK?)

Compute audio tags for the given stream.

param stream

An OfflineStream.

param topK

Number of top results (number, default -1 for all).

returns

An array of AudioEvent objects, each with:

• name (string) — Event category name.

• prob (number) — Probability score.

• index (number) — Category index.

Properties

• tagger.config — The configuration object.

Example

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

const tagger = new sherpa_onnx.AudioTagging({
  model: { ced: './ced.onnx' },
  labels: './labels.txt',
  topK: 5,
});

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

const events = tagger.compute(stream);
for (const event of events) {
  console.log(`${event.name}: ${event.prob.toFixed(3)}`);
}

Notes

• The input audio should be mono, 16kHz, float32 in [-1, 1].

• Use topK to limit the number of returned results.