Skip to main content

Playback and queues

Browser helper​

import {KittenTTS, createBrowserAudioPlayer} from '@kittentts/web';

const tts = await KittenTTS.create({
  player: createBrowserAudioPlayer(),
});

await tts.speak('This plays through an HTML audio element.');

Browser autoplay policies apply. Call playback from a user interaction when required.

Generate, then play​

const result = await tts.generate('Prepare the interface first.');

await tts.play(result, {
  onPlaybackStart: () => startHighlighting(result.wordTimings),
});

FIFO queue​

const queue = tts.createPlaybackQueue();

await queue.enqueue(firstResult);
await queue.enqueueText('Generate after earlier clips.', {voice: 'luna'});

queue.clear() rejects pending tasks. queue.stop() clears pending tasks and stops active audio. Queue properties expose total length, pending count, and active state.

Custom player​

const player: AudioPlayer = {
  async play(samples, sampleRate, options) {
    options?.onPlaybackStart?.();
    await sendToAudioEngine(samples, sampleRate);
  },
  async stop() {},
};

Use a custom adapter for Web Audio, Electron, game engines, server pipelines, or framework-owned media components. Resolve play() after audio finishes when strict queue sequencing is required.

Last updated on