Speech

Convert text to speech on device with Apple’s AVSpeechSynthesizer. The audio is returned as WAV. Personal Voice needs iOS 17.

Usage

import { apple } from '@react-native-ai/apple'
import { generateSpeech } from 'ai'

const { audio } = await generateSpeech({
  model: apple.speechModel(),
  text: 'Hello from Apple on-device speech!',
})

console.log(audio.uint8Array)
console.log(audio.base64)

The sample rate and bit depth depend on the voice, commonly 44.1 kHz 16-bit PCM. Some voices return 32-bit float, which the WAV header reflects.

Voice and Language

Pass language to use the system’s default voice for that locale:

await generateSpeech({
  model: apple.speechModel(),
  text: 'Bonjour tout le monde!',
  language: 'fr-FR',
})

Or pick a specific voice by its identifier, which takes priority over language:

import { AppleSpeech } from '@react-native-ai/apple'

const voices = await AppleSpeech.getVoices()

await generateSpeech({
  model: apple.speechModel(),
  text: 'Custom voice synthesis',
  voice: 'com.apple.voice.super-compact.en-US.Samantha',
})

Each voice has:

• identifier and name
• language, a BCP 47 code such as en-US
• quality: default, enhanced, or premium
• isPersonalVoice and isNoveltyVoice (iOS 17+)

The system manages voices. Enhanced and premium voices may need a one-time download, which users start in Settings → Accessibility → Read & Speak → Voices.

Speed

speed multiplies the voice’s default rate, so 2 speaks twice as fast and 0.5 half as fast. Apple caps the rate, so extreme values are clamped:

await generateSpeech({
  model: apple.speechModel(),
  text: 'Speaking a little faster.',
  speed: 1.25,
})

Personal Voice

On iOS 17+, a Personal Voice the user created is only listed once they let your app use it. Ask when your app needs it, for example when it opens a voice picker. iOS shows the prompt once, and later calls return the saved answer:

const status = await AppleSpeech.requestPersonalVoiceAuthorization()
// 'authorized' | 'denied' | 'notDetermined' | 'unsupported'

if (status === 'authorized') {
  const personalVoices = (await AppleSpeech.getVoices()).filter(
    (voice) => voice.isPersonalVoice
  )
}

Right after the user allows access, iOS can take a moment to add the Personal Voice to the list. If it is missing, call getVoices() again shortly after, for example when your voice picker next appears.

Direct API Access

AppleSpeech is the native module behind the model:

import { AppleSpeech } from '@react-native-ai/apple'

const voices = await AppleSpeech.getVoices()

// WAV bytes as an ArrayBuffer
const buffer = await AppleSpeech.generate('Hello world', {
  language: 'en-US',
})