Apple introduces Audio Intelligence for Apple Watch Series 12 and Ultra 4, featuring four opt-in capabilities: Sound Recognition, automatic Music Recognition with Shazam, Live Rewind, and Siri Recap. These features leverage on-device microphones and Apple silicon to detect sounds, identify music, and provide contextual summaries. Privacy is central, with all processing kept on-device and user-controlled.

Live Rewind captures the last 15 seconds of speech after a deliberate Digital Crown double-press, storing audio in a secure buffer on the S11 chip. It does not transcribe continuously or save data unless the user chooses. Siri Recap generates periodic summaries based on user-set schedules, omitting sensitive content and allowing easy toggling via the Watch app or Control Center.

Apple's 11-page privacy paper details that audio data never leaves the device, and the S11 chip's Secure Exclave isolates it from the OS and apps. Users can enable or disable each feature at any time, ensuring full control. The features roll out in beta later this year with English support, with more details at launch.