Catch missed words as Apple Watch AI enters beta

Apple’s Audio Intelligence features are beginning beta testing for Apple Watch, with tools for sound and song recognition, conversation rewind, and AI recaps. Live Rewind shows a recent 15-second snippet, while Siri Recaps creates short notes with controls for reviewing and deleting them.

WTF Index TERMINATOR
◄ Terminator 2 Idiocracy 1 ►

Conversation listening raises privacy concerns, though the features are opt-in and include visible recording cues and user controls.

Catch missed words as Apple Watch AI enters beta

Apple is beginning beta testing a set of opt-in Audio Intelligence features for Apple Watch. The tools are designed to help wearers recognize sounds and songs, revisit a brief stretch of conversation, and create concise notes from discussions.

Four features, including two that listen to conversations

Audio Intelligence includes sound recognition for cues such as sirens, alarms, and doorbells, which may help deaf and hard-of-hearing users identify them. Song recognition uses Apple’s Shazam app to identify music playing nearby, including the song name and artist.

The two conversation-focused tools are Live Rewind and Siri Recaps. Both involve listening to speech, a capability that raises privacy questions amid backlash to AI recording devices such as Meta’s AI glasses and other AI wearables.

Apple says the features are opt-in and describes design choices intended to make their use visible and limit how audio and notes are handled. Their availability is beginning with developer beta testing in the iOS 27.3 update, after the features were announced at the company’s Surprise and Shine hardware event in September.

Live Rewind brings back the last 15 seconds

Live Rewind is meant for moments when a wearer misses something just said. Double-pressing the Digital Crown displays a text snippet of the previous 15 seconds on the watch face. That could be useful for catching verbal directions or recalling a long list of restaurant specials.

The wearer can read the snippet and leave it unsaved, or save it to the standalone Siri app that shipped with iOS 27. The saved text could serve as a quick reminder of a name or a note from a fast-paced presentation. An “Ask Siri” option below the snippet can also answer questions about a term or option mentioned in the conversation.

The interaction is signaled with an audible chime and an animation on the watch. Apple says the gesture, sound, and visual display are intended to let people nearby know that something said has been recorded. Unsaved snippets are deleted automatically after 30 seconds.

Siri Recaps turns conversations into brief notes

Siri Recaps produces a short sentence and bullet points highlighting key information from a conversation. Apple says the feature does not identify speakers, retain raw audio, or provide word-for-word transcripts. Depending on the discussion, the notes may include dates, deadlines, prices, payment-plan details, or other factual points.

Apple suggests the feature could help with work meetings, parent-teacher conferences, and personal training sessions, especially when someone cannot take notes while listening. It is intended to capture tasks and information without adding emotion or tone. Recaps will not record conversations while the wearer is listening to music with headphones or using their noise cancellation features.

When enabled, Siri Recaps prompts the wearer to review saved recaps at the end of the day and again the first time they unlock the watch in the morning. Notes can be saved to the Notes app or exported to the Journal app. If the wearer takes no action, recaps delete themselves within seven days.

Wearers can turn the feature on manually or set a schedule, such as limiting it to work hours or preventing it from running at night. It can also be switched off at any time through Control Center. Apple says Siri Recaps is designed to omit sensitive information, including government-assigned identifiers, financial and authentication data, certain personal identifiers, and language that would promote self-harm.

Apple describes a multi-device processing path

The features for Apple Watch Series 12 and Apple Watch Ultra 4 depend on the S11 chip, the built-in microphone, Apple Intelligence models running on the iPhone, and Apple’s Private Cloud Compute. Apple describes a secure enclave as a hardware-protected memory space, separate from the rest of the operating system, where a dedicated audio buffer processes raw audio.

Lightweight models on the watch detect whether a conversation is happening that includes the wearer. The audio is then transferred to the paired iPhone through a secure pairing mechanism. On-device processing condenses the speech into text with filler words removed, and that text is encrypted before being sent to Private Cloud Compute to create a recap.

Apple says the recap and the summaries used to create it are deleted from Private Cloud Compute immediately afterward. Recaps available on the iPhone are accessible only to the user, and are encrypted when synced across Apple devices. Live Rewind snippets also transfer to the iPhone for processing.

Apple says the features should not affect watch or phone battery life. After developer beta testing, they are expected to move to a public beta for broader testing before becoming publicly available.