Skip to main content

Generation and streaming

Generate one result​

const result = await tts.generate('Save this as audio.', {
  voice: 'luna',
  speed: 1.05,
});

console.log(result.samples);       // Float32Array
console.log(result.sampleRate);    // 24000
console.log(result.duration);
console.log(result.wordTimings);

Encode output​

const wav = result.wavData();
const wavBase64 = result.wavBase64();
const mp3 = await result.mp3Data({bitRate: 128});
const mp3Base64 = await result.mp3Base64({bitRate: 128});

Stream long input​

for await (const chunk of tts.stream(articleText, {
  voice: 'jasper',
  speed: 1.0,
})) {
  await tts.play(chunk);
}

The SDK divides long text into sentence groups and yields a complete result after each local inference. This improves time to first audio and provides UI update boundaries.

Resource lifecycle​

Reuse the initialized instance. Call dispose() when the application is permanently finished with it. Analytics can be disabled with analytics: false; no input text or generated audio is included in the anonymous generation metadata.

Last updated on