Add speech-to-text features to your React Native app using Apple’s SpeechAnalyzer and SpeechTranscriber APIs in iOS 26. Runs entirely on-device, with system-managed language models and support for multiple languages.

Callstack Blog

Callstack introduces on-device speech transcription for React Native apps using Apple's new SpeechAnalyzer and SpeechTranscriber APIs from iOS 26, integrated with the AI SDK's experimental_transcribe function. The implementation runs entirely on-device with no cloud dependency, supports multiple languages via ISO 639-1 codes, and handles language model asset management automatically without impacting app bundle size. Performance benchmarks show it transcribes audio over 2.2x faster than Whisper Large V3. The API accepts ArrayBuffer or base64 audio and returns transcribed text, word-level segments with timestamps, and total duration.

On-Device Speech Transcription with Apple SpeechAnalyzer and AI SDK

Under the Hood: Apple's SpeechAnalyzer Framework