Apple has started beta testing its Audio Intelligence suite for the Apple Watch in the developer build of iOS 27.3, the company announced. The set of four opt-in features is designed for Apple Watch Series 12 wearers (with system support described for Apple Watch Ultra 4) and includes sound recognition, song identification via Shazam, Live Rewind, and Siri Recaps.
How Live Rewind and song detection work
Live Rewind lets a wearer double-press the digital crown to display a text snippet of up to the last 15 seconds of speech, in case something was missed. When invoked, the watch plays a chime and shows an animation intended to signal to nearby people that a short capture occurred. Snippets disappear after 30 seconds unless the user saves them to the standalone Siri app on the paired iPhone, and an “Ask Siri” option can appear beneath a snippet for follow-up questions.
The song recognition feature taps Apple’s Shazam technology to identify background music and present the song title and artist on the watch.
Siri Recaps, privacy and controls
Siri Recaps summarizes conversations into a short sentence with bullet points highlighting factual items such as dates, pricing, or deadlines. Apple says the feature does not identify speakers, save raw audio, or produce word-for-word transcripts. Recaps are intended to capture key information without adding emotion or tone.
Apple notes several user controls: Siri Recaps can be enabled manually or scheduled (for example, only at work), paused from Control Center, and will not run when the user is listening to music with headphones or using noise cancellation. Saved recaps can be exported to the Notes app or Journal app; unsaved recaps delete themselves after seven days. The system prompts users to review recaps after first unlocking the watch each morning when the feature is active.
On-device and cloud processing, Apple says, are designed to protect privacy. Lightweight models on the S11 chip detect when the wearer is part of a conversation; a secure enclave processes raw audio and transfers only condensed text (with filler words removed) to the paired iPhone. That condensed text is encrypted and sent to Apple’s Private Cloud Compute (PCC) to generate the recap, and Apple says those interim inputs are deleted immediately from PCC. Recaps are encrypted for the user across their devices. Apple also states these features should not affect watch or phone battery life.
The Audio Intelligence features are in developer beta now and will expand to the public beta before a wider release.
Original source: TechCrunch AI