Apple has launched developer beta testing for new artificial intelligence tools on the Apple Watch, introducing ambient conversation tracking and audio recognition features through the iOS 27.3 update.
Apple Deploys Audio Tools in iOS 27.3
The beta release brings four opt-in capabilities to the Apple Watch Series 12 and Apple Watch Ultra 4. First announced at Apple’s September hardware event, the Audio Intelligence suite includes sound recognition, song identification via Shazam, and two conversation-monitoring features: Live Rewind and Siri Recaps.
Apple confirmed that the features will not initially be available in the EU. Initial availability is restricted to English-language users. Hardware requirements are strict. The tools require the S11 chip found exclusively in the newest watch models, paired with an iPhone 16 or later running iOS 27.3, explicitly excluding the iPhone 16e.
The features officially launched on November 2, 2026 for developer beta testing. Alongside Live Rewind and Siri Recap are Sound Recognition, which can notify wearers about sounds such as sirens and doorbells, and automatic Music Recognition with Shazam.
Display Speech Text with Live Rewind
Live Rewind serves as a short-term buffer, displaying the previous 15 seconds of speech as text when a user double-presses the Digital Crown. The two conversation tools approach memory retention differently. The action triggers an audible chime and visual animation to alert people nearby. Unsaved text snippets vanish after 30 seconds.
Live Rewind is activated only by a deliberate double-press of the Digital Crown. Missed what the waiter just listed, or the directions someone rattled off? Should anyone have queries regarding past remarks, an “Ask Siri” prompt is also available beneath the transcribed text.
Siri Recaps operates on a larger scale, running in the background to generate titles and bullet-point summaries of meetings or appointments. Users can activate the feature manually or establish schedules, such as limiting recapping to work hours. Users can activate it manually or set a timetable—such as limiting it strictly to working hours while excluding nights—and may disable it via Control Center at any moment. Unsaved notes auto-delete after seven days.
not accessible to Apple
Apple, Technical Support Documentation
Apple cautions that the automated summaries may be incomplete or inaccurate, warning users against treating the generated points as authoritative records for critical details like medical instructions. Apple notes that the generated snippets and summaries might prove inaccurate or incomplete, and warns that crucial information could be left out, misunderstood, or incorrectly condensed.
Process Raw Audio on Secure Enclave
Raw audio is processed within a secure enclave on the watch’s S11 chip—a hardware-isolated memory space inaccessible to watchOS, applications, and Apple itself. To address privacy concerns surrounding ambient listening, Apple employs a multi-layered security model. The S11 chip processes sensor data separately from the rest of the system. Lightweight models detect active conversations, and encrypted audio moves to the paired iPhone, where filler words are removed.
The features require pairing with an iPhone running compatible Apple Intelligence models. The compressed text then travels to Apple’s Private Cloud Compute infrastructure for final summarization before being immediately deleted from servers. The system is programmed to filter out sensitive details, including government identifiers, financial data, authentication credentials, and language associated with self-harm.
Apple Defends Revenue Strategy and AI Timeline
In an interview with Wired’s Steven Levy, John Ternus, who took over in September, pushed back against perceptions that the company trails competitors in AI after delays to its upgraded Siri, which arrived this summer, stating that Apple refuses to ship features simply for their own sake. The rollout coincides with executive defense of Apple’s artificial intelligence timeline. Meanwhile, CFO Kevan Parekh pointed to AI as a future driver for device upgrades and new revenue streams across first-party services and third-party partnerships.
