Apple Watch Series 12 gets AI to summarize conversations

Apple's support page for the Apple Watch Series 12 and Apple Watch Ultra 4 introduces four audio features under the banner "Audio Intelligence," including one that captures the last 15 seconds of speech around the wearer as text and another that summarizes their conversations across the day.

The page lists four capabilities. Sound Recognition flags sirens, alarms and doorbells and, per Apple, notifies the wearer even without an iPhone nearby, which the company frames as an accessibility feature for deaf and hard of hearing users. Music Recognition uses Shazam to identify songs playing nearby and drops the title and artist into the Smart Stack without user interaction. Live Rewind, activated by double-pressing the Digital Crown, recalls the previous 15 seconds of speech as text snippets that the wearer can query with Siri or save to the Siri app. Siri Recap, in Apple's own phrasing, "Takes notes on your conversations throughout your day and creates high-level Apple Intelligence–generated summaries" that auto-delete after seven days unless the user saves them. Three tracked experts in our Who's Who directory shared the support page this week.

Apple attributes local processing to a new S11 chip with a Secure Exclave, described on the page as a hardware-isolated compartment, and routes heavier inference to Apple Foundation Models via Private Cloud Compute. Song matching runs off acoustic signatures rather than raw audio.

Sound Recognition and Music Recognition require an iPhone 11 or later, or an iPhone SE second generation or later, running iOS 27. Live Rewind and Siri Recap are labeled beta, need an iPhone 16, 17 or 18 series device, ship in English only at start, and are gated to users aged 13 and up with parental approval required between ages 13 and 17. Apple says both will launch later in 2026.