speech-recognition

Transcribe live microphone audio to text on iOS by routing AVAudioEngine buffers to SpeechAnalyzer or SFSpeechRecognizer. This streams asynchronous, progressive results with time-indexed transcripts while managing microphone and speech authorization.

On-device speech recognition works on iOS using SpeechAnalyzer or SFSpeechRecognizer, providing offline transcription without server dependency. Server-backed recognition offers broader locale support, while on-device processing ensures privacy and consistent availability.

Transcribe pre-recorded audio files using SpeechTranscriber or SFSpeechRecognizer to process the audio asset. The framework supports configurable locale handling and optional alternatives, returning the finalized text transcript for immediate use.

Handle microphone and speech recognition authorization by explicitly requesting user permissions before starting iOS transcription. Robust authorization management ensures speech access is granted, allowing SpeechAnalyzer to capture live audio buffers without interruption.

Install required model assets using AssetInventory before using SpeechAnalyzer for speech to text. Managing these assets ensures the necessary on-device recognition models are downloaded and available before initiating analyzer sessions.

iOS speech recognition sessions require explicit finalization or cancellation to properly conclude analyzer input streams. Finalizing returns the complete transcription result, while canceling halts processing and releases the active session resources.