Apple Watch Series 12 Launches Audio Intelligence Beta for Conversation Rewind and Summaries

Apple has begun beta testing its Audio Intelligence suite on Apple Watch Series 12, delivered through iOS 27.3 for developers. The suite comprises four opt‑in features: sound detection, song identification, Live Rewind, and Siri Recaps.

Live Rewind lets a wearer double‑press the Digital Crown to retrieve a 15‑second text snippet of the most recent speech. The snippet appears with an audible chime and animation, then auto‑deletes after 30 seconds unless saved to the new standalone Siri app.

Siri Recaps goes further by generating concise, bullet‑point summaries of entire conversations without storing raw audio or speaker IDs. Apple positions the feature for meetings, training sessions, and other scenarios where note‑taking is impractical.

Why it matters for GPU and AI infrastructure

For AI infrastructure providers, the rollout signals growing demand for on‑device, privacy‑first inference engines that can process continuous audio streams in real time. Chips optimized for low‑latency speech‑to‑text and summarization will become a differentiator in the wearable market.