Tools

AI-generated text

Apple begins developer beta of Audio Intelligence for Apple Watch Series 12 with Live Rewind and Siri Recaps

Apple has started a developer beta of its Audio Intelligence suite for Apple Watch Series 12 and Apple Watch Ultra 4 in iOS 27.3, introducing four opt-in features including sound and song recognition, Live Rewind, and Siri Recaps.

Apple begins developer beta of Audio Intelligence for Apple Watch Series 12 with Live Rewind and Siri Recaps

Apple has begun rolling out the Audio Intelligence features in a developer beta of iOS 27.3 for Apple Watch Series 12 and Apple Watch Ultra 4. The suite, first announced at Apple’s Surprise and Shine hardware event in September, includes four opt-in AI features: sound recognition, song recognition, Live Rewind and Siri Recaps.

What the features do

  • Sound recognition: helps deaf and hard-of-hearing users identify important sounds such as sirens, alarms or doorbells directly from the watch.
  • Song recognition: uses the Shazam app to identify music playing in the background and return the song title and artist.

The two features that listen to human conversation — Live Rewind and Siri Recaps — are likely to attract the most attention.

Live Rewind — replay recent speech

Live Rewind lets the wearer go back a short time to see what was just said. If you miss spoken directions or a long verbal list, a double-press of the watch’s digital crown shows a text snippet of the last 15 seconds on the watch face. You can view the snippet, which disappears permanently if you do nothing, or save it to the standalone Siri app introduced in iOS 27 for later access.

When the digital crown is double-pressed, an audible chime and an on-screen animation play to notify nearby people that something was captured. Unsaved 15-second snippets are automatically deleted after 30 seconds.

An “Ask Siri” option can appear below the snippet to let you query the captured content (for example, to clarify a technical term or a menu detail mentioned by a server).

Siri Recaps — high-level conversation summaries

Siri Recaps generates brief notes about conversations you participate in, but according to Apple it does not identify speakers, does not save raw audio, and does not provide verbatim transcriptions. Instead, the feature produces a short summary sentence and a few bullet points highlighting key facts such as important dates, deadlines, pricing or other factual items.

Apple suggests use cases like work meetings, parent-teacher conferences or personal training sessions where you might need to remember tasks or facts but cannot take notes. Recaps do not run while you are listening to music through headphones with noise cancellation.

When enabled, you are prompted to review saved Siri Recaps and decide which, if any, to keep. Saved notes can be moved to the Notes app or exported to a Journal app for more personal content. If you do nothing, recaps are deleted after seven days. You’ll also see a reminder to review recaps when you first unlock your Apple Watch in the morning.

Siri Recaps can be turned on manually or scheduled (for example, only at work or never at night). Even when scheduled, recapping can be turned off at any time from Control Center.

Apple says the feature is designed to omit sensitive information such as government-assigned identifiers, financial and authentication data, certain personal identifiers, and language that promotes or details self-harm.

How Apple says it protects data

Audio Intelligence relies on the S11 chip, the watch’s built-in microphone, Apple Intelligence models that run on the paired iPhone, and Apple Private Cloud Compute (PCC). The always-on processor includes a Secure Enclave — a hardware-protected memory area inaccessible to the rest of the operating system — where a dedicated audio buffer processes raw audio so it is not accessible to anyone, including Apple. Audio is accessible only to the paired iPhone via a secure pairing mechanism.

Lightweight AI models on the S11 detect whether a conversation involving the wearer is occurring and then transfer data to the iPhone’s secure enclave. The iPhone performs on-device processing to convert the audio into condensed text (removing filler words). This condensed text is encrypted and sent to Apple Private Cloud Compute to generate the conversation recap; Apple says the recap and any intermediate data used to create it are deleted from PCC immediately afterwards.

The recaps available on the iPhone are accessible only to the user and are encrypted when synced across a user’s Apple devices. Live Rewind also transfers audio to the iPhone for processing to create the 15-second snippets.

Apple says these features should not materially affect the battery life of the watch or phone. After the developer beta, the company plans broader public beta testing before making the features available to all users.