AuralKit 3.0 adds video track transcription

Apple's SpeechAnalyzer framework delivers capable on-device transcription, but managing AVFoundation buffer conversion and session lifecycles remains a notorious headache for Swift developers. AuralKit abstracts the underlying audio graph plumbing to make native offline speech recognition practical for complex media and screen capture workflows.

• –Dedicated media track transcription allows apps to isolate commentary or localized audio streams without manual demuxing or losing timeline synchronization.
• –Overhauled lifecycle semantics replace ambiguous stop calls with explicit finish and cancel methods, ensuring buffered audio drains properly upon completion.
• –ScreenCaptureKit pipeline integration feeds system and app audio directly into SpeechAnalyzer, simplifying local meeting notes and screen recording transcription.
• –On-device processing leverages Apple's native speech models for zero-latency, private transcription without third-party cloud API costs.