Apple Opens Audio Intelligence Beta to Developers for Apple Watch

Apple has begun beta testing Audio Intelligence, its AI-powered audio feature set for Apple Watch, with developers as part of the iOS 27.3 update, according to TechCrunch. Apple first announced the feature set at its Surprise and Shine hardware event in September.

Audio Intelligence bundles four opt-in features for Apple Watch Series 12 wearers: detection of important sounds, song identification, catching up on conversations, and note-taking. The sound recognition function is intended to help deaf and hard-of-hearing users identify sounds such as sirens, alarms, and doorbells. The song recognition function uses Shazam to detect background music and report the song name and artist.

The conversation feature drawing the most attention is Live Rewind. A double-press of the digital crown, the dial on the side of the watch, shows what was said in the last 15 seconds as a text snippet on the watch face. An audible chime and an on-watch animation accompany it. An Ask Siri option sits below the snippet for asking questions about what was said. Snippets are deleted automatically after 30 seconds unless the user saves them, and saved snippets can be kept in the standalone Siri app that shipped with iOS 27 for later access.

A companion feature, Siri Recaps, summarizes conversations with a short sentence and bullet points identifying key points. According to the report, Recaps does not identify speakers, does not save raw audio, and does not provide word-for-word transcriptions.

Live Rewind is enabled from the iPhone. Users open the Watch app, tap Siri, then tap Live Rewind, according to Apple's support documentation.

In my view, the retention settings deserve the closest reading. A 15-second look-back with a 30-second auto-delete window, and a summary format that discards speaker identity and raw audio, narrows what the device keeps. That narrowing is a design decision, and it shapes how far a wrist device can be used to record other people in a room. Whether those constraints satisfy users, bystanders, and regulators is a question the announcement does not settle.

The broader context here is accessibility. Routing siren, alarm, and doorbell detection to the wrist, with a vibration and a visual signal, gives deaf and hard-of-hearing users a route to environmental awareness that does not depend on hearing. Shazam-based song recognition and conversation recaps are more conventional consumer features, but they run on the same on-device and cloud-assisted pipeline that makes the accessibility functions possible.

The announcement is a beta for developers, not a general release. Developers will be the first to test how the rewind buffer, the Siri app handoff, and the summary output behave across real conversations. Their findings will shape how the features reach ordinary owners of the Series 12. For this audience, the open questions are latency on the watch, how much processing stays local, and how saved snippets are governed once they leave the 30-second window. Those answers will matter more to the feature's future than the feature list does.

In my view, the most useful outcome is a set of assistive functions that people with hearing loss can rely on day to day. The conversation features are a secondary benefit, and they will be judged by how well they respect the people nearby who never opted in.