On-device speech synthesis is now available in the AI SDK for React Native. Powered by Apple’s AVSpeechSynthesizer, it delivers private, low-latency text-to-speech with rich voice options and offline support.

Callstack Blog

Callstack has added on-device text-to-speech synthesis to the AI SDK for React Native via a new Apple provider. Powered by Apple's AVSpeechSynthesizer, it processes audio entirely on-device using raw PCM buffers to produce WAV output with no disk I/O. The API integrates with Vercel's AI SDK, supports voice selection by identifier or language, provides access to Personal Voice on iOS 17+, and works offline with no data leaving the device. This completes the Apple provider feature set alongside generation, embeddings, and transcription.

On-Device Text To Speech on Apple Devices with AI SDK