Apple Beta Tests 'Audio Intelligence' for Watch Series 12

Apple has begun beta testing "Audio Intelligence," a suite of AI-powered features for the Apple Watch Series 12 and Apple Watch Ultra 4, as TechCrunch first reported. Initially rolling out to developers via the iOS 27.3 update, the technology was first introduced during the company's Surprise and Shine hardware event in September.

The update introduces four opt-in features. Two are focused on environmental sounds: one allows deaf and hard-of-hearing users to identify sounds such as doorbells, alarms, or sirens, and another integrates with the Shazam app to identify background music and artists.

However, the more contentious additions involve the recording of human speech through "Live Rewind" and "Siri Recaps." TechCrunch reports that Live Rewind allows users to double-press the digital crown to view a text snippet of the previous 15 seconds of conversation. This is intended for situations such as receiving verbal directions or listening to restaurant specials. To notify others that recording has occurred, Apple has implemented a combination of a visual animation, a gesture, and an audible chime. These snippets are deleted after 30 seconds unless the user saves them to the Siri app provided in iOS 27. Users can also select an "Ask Siri" option to clarify technical terms or other details mentioned in the snippet.

Siri Recaps expands this ambient listening by generating high-level summaries of conversations. TechCrunch notes that Apple intends for this to be used during parent-teacher conferences, work meetings, or personal training sessions. Rather than providing word-for-word transcriptions or identifying specific speakers, the feature produces a short sentence and bullet points highlighting factual data, such as payment plans, deadlines, or pricing.

Apple states that Siri Recaps is designed to omit authentication data, financial information, government-assigned identifiers, and language promoting self-harm. Users can enable the feature manually or set a schedule—such as only during work hours—and can disable it via the Control Center. Recaps are deleted after seven days unless the user saves them to the Journal or Notes apps. The system is designed to stop recording if the user is utilizing noise cancellation or listening to music through headphones.

Technically, the system relies on the S11 chip, the built-in microphone, and Apple Intelligence models on the iPhone, supported by Apple's Private Cloud Compute. TechCrunch reports that raw audio is processed within a secure enclave—a hardware-protected memory space—to ensure the audio is not accessible to Apple or other parts of the operating system. Lightweight AI models on the S11 chip detect conversations, which are then transferred to the iPhone's secure enclave. The iPhone converts audio into "condensed text" (removing filler words), which is then encrypted and sent to Private Cloud Compute to generate the final recap.