Streaming TTS - RunAnywhere Documentation

synthesizeStream emits AudioChunk values as they are produced, so playback starts on the first chunk instead of the last.

for await (const chunk of RunAnywhere.tts.synthesizeStream(longText)) {
  player.enqueue(chunk)
}

Signature

synthesizeStream(text: string, options?: TtsOptions): AsyncIterable<AudioChunk>

When it is worth it

For a sentence, speak is simpler and the difference is invisible. For a paragraph, or for a model reply being streamed as it generates, streaming synthesis is what keeps the gap between “the model finished” and “the user hears something” short.

Pairing it with a streaming completion

Speak each sentence as the model finishes it:

let sentence = ''

for await (const event of RunAnywhere.llm.generateStream(prompt)) {
  if (event.type !== 'textDelta') continue
  sentence += event.text

  if (/[.!?]/.test(event.text)) {
    RunAnywhere.tts.speak(sentence)
    sentence = ''
  }
}

if (sentence) RunAnywhere.tts.speak(sentence)

For a full conversation rather than one reply, use the voice session, which handles turn-taking and interruption.

AudioPlaybackManager

@runanywhere/core exports AudioPlaybackManager, which wraps the native playback path the voice session uses. Prefer it over hand-rolling a player from chunks: it already handles the audio session and the scheduling.

import { AudioPlaybackManager } from '@runanywhere/core'

Stopping

await RunAnywhere.tts.stop()