Apple Adds Opt-In Conversation Recap and Audio Rewind to Apple Watch Developer Beta

Apple has rolled out developer beta access for Siri Recap and Live Rewind, two opt-in Audio Intelligence tools designed to summarize spoken conversations and transcribe recent audio on its smartwatches, ZDNET reported. The features are available on the Apple Watch Series 12 and Apple Watch Ultra 4 when paired with an iPhone running the iOS 27.3 developer beta.

Apple initially previewed both tools in September alongside the Series 12 and Ultra 4 announcements, but held back software availability until the current test release. The deployment introduces persistent listening mechanics to Apple's wearable line, paired with on-chip hardware boundaries intended to mitigate the privacy risks of ambient audio capture.

Siri Recap generates concise outlines of spoken discussions rather than verbatim transcripts, focusing on action items, dates, times, and prices. In a demonstration documented by ZDNET, the software captured key planning points from an informal conversation between a parent and child regarding college admissions. Apple configured the tool to omit speaker names unless participants formally introduce themselves or are repeatedly addressed by name. The system also filters out personal identifiers, financial details, authentication credentials, and references to self-harm.

Summaries appear in a dedicated Recaps tab within the Siri application and can be manually exported to Apple Notes or Journal. Recaps that are not saved are automatically deleted after seven days. Users can toggle the tool through Control Center, schedule automated activation based on arrival at locations such as offices or lecture halls, and receive daily status notifications. The recording system pauses automatically during phone calls, FaceTime sessions, audio playback, or when connected to Bluetooth accessories like AirPods.

The second feature, Live Rewind, maintains a local audio buffer to recover missed dialogue. Double-pressing the Digital Crown chimes, shows an on-screen animation, and presents a transcript of the preceding 15 seconds of conversation, which remains visible for 30 seconds. Wearers can save the transcript, dismiss it, or forward the snippet to Siri for reference queries. Untranscribed audio is continuously overwritten and cleared locally without being transmitted to Apple servers, while saved excerpts are encrypted in iCloud.

To address security concerns over persistent microphone access, Apple isolates raw audio within a dedicated Secure Exclave on the wearable's S11 processor. When the S11 chip detects an active conversation, audio enters the Secure Exclave, is encrypted, and transmits across a secure link to the paired iPhone's Secure Exclave. The handset condenses the audio into edited text before routing it to Apple's Private Cloud Compute infrastructure to compile the final summary, preventing Apple or outside parties from accessing the raw audio stream.

Both capabilities are disabled by default during device setup and require manual opt-in through the iPhone's Watch companion app. Apple has not yet announced a timeline for general consumer availability.