Apple introduced Apple Watch Series 12 on September 9, 2026, with two announced Audio Intelligence features aimed at helping users recall recent conversations. Apple says Live Rewind can show a text snippet from the previous 15 seconds of conversation, while Siri Recap can generate a title and key points for later review.
Apple says both features will launch in beta later in 2026. They will require an Apple Intelligence-enabled iPhone 16 or later, excluding iPhone 16e, and will initially support English but not be available in the European Union. Apple also says the features are opt-in and do not create or store audio recordings.
Contents
What Apple announced
Live Rewind is activated with a double press of the Apple Watch Digital Crown. The watch then displays a text snippet covering the previous 15 seconds of conversation. Users can ask Siri about the snippet or save it to the Siri app for later review.
Siri Recap is a separate feature. When enabled, Apple says it can produce a brief title and a set of key points from a conversation in the Siri app. It is intended as a high-level summary rather than a complete transcript.
That distinction matters. A transcript attempts to preserve spoken words, while a summary compresses and paraphrases information. Siri Recap may therefore be useful for recovering the general subject or main points of a discussion, but Apple has not presented it as a verbatim record.
Apple says Live Rewind provides a chime, a full-display animation and a microphone indicator when activated. These signals are intended to tell people nearby that the feature is being used.
The announcement also describes Sound Recognition, which can identify sounds such as sirens, alarms, doorbells and a baby crying, then notify users on the watch even when their iPhone is not nearby. Apple positions this as an on-device intelligence feature for people who are deaf or hard of hearing. It is distinct from the new conversation-recall and summary functions.
How the features are meant to work
Apple says the new Audio Intelligence functions combine the Apple Watch Series 12's S11 chip and built-in microphone with Apple Intelligence models on the iPhone and Private Cloud Compute. The company has not explained the precise division of processing between the watch, the paired iPhone and cloud infrastructure.
A microphone converts sound pressure into an electrical signal that hardware or software can analyse. That makes it possible for a device to detect sound patterns or process speech without necessarily retaining a conventional audio recording. Apple says the raw audio used for these features is processed in a hardware-isolated part of the S11 chip called Secure Exclave.
According to Apple, Secure Exclave processes the audio separately and immediately deletes it. The company also says Audio Intelligence features do not identify or attribute speakers. That may limit their usefulness in conversations involving several people, but it also means the system is not designed to label who said particular words.
Apple says Live Rewind and Siri Recap are available only after users opt in to each feature. Siri Recap can be switched on or off from Control Center.
The architecture described by Apple does not mean every part of the system runs entirely on the watch. Apple specifically says that some Audio Intelligence functions relying on server-side models, including Live Rewind and Siri Recap, are subject to daily usage limits. The release does not state the size of those limits.
Privacy design and its limits
Apple says Audio Intelligence features do not create or store audio recordings. It further says raw audio is inaccessible to the operating system, apps, the user and Apple itself.
The company says the resulting Live Rewind snippets and Siri Recap summaries are end-to-end encrypted in the Siri app, with iCloud syncing, and cannot be accessed even by Apple. End-to-end encryption is designed to restrict readable access to the intended endpoints, although the strength of that protection depends on the implementation and key-management design.
These are manufacturer claims rather than independently verified findings. The announcement does not provide an audit, technical security review or independent accuracy benchmark. It also does not explain how often audio is captured or processed before a user activates Live Rewind, beyond saying that raw audio is not stored.
The privacy signals are particularly relevant because a wrist-worn device can be present during ordinary conversations. The chime, animation and microphone indicator make activation more visible, but the release does not provide a legal analysis of recording or consent rules in different jurisdictions.
The features' real-world accuracy and failure modes remain unmeasured in the supplied evidence. Apple has not published error rates or testing results for accents, background noise, overlapping speakers or technical terminology. No study design, sample size or comparator is provided.
Availability and what remains unknown
Apple Watch Series 12 uses the S11 chip and a new Health Sensing System. Apple says Audio Intelligence is also available on Apple Watch Ultra 4, subject to the same stated ecosystem requirements. The features require watchOS 27 and a compatible Apple Intelligence-enabled iPhone 16 or later, excluding iPhone 16e.
Live Rewind and Siri Recap are scheduled to arrive in beta later in 2026, but Apple has not given an exact date. The initial release will be English-only and will not initially be available in the European Union. Apple also says not all features are available on all devices or in all regions.
In India, Apple Watch Series 12 can be pre-ordered from September 9, with store availability beginning September 18. The announcement does not provide an India-specific price or confirm that Live Rewind and Siri Recap will be available in India at the beta launch.
The most important next evidence will be independent testing of the features in real conversations, clearer documentation of where processing occurs and more detail about retention and deletion. It will also be important to see how daily server-side limits affect practical use and whether Apple expands language and regional support.