Apple has plenty of shiny numbers to sell the Apple Watch Series 12 and Apple Watch Ultra 4 with. The new Health Sensing System, the Readiness feature, and more frequent HRV readings are just a couple of things that make these a meaningful upgrade. This is especially true for anyone who treats an Apple Watch primarily as a health and fitness device.
Though the feature that caught my eye has almost nothing to do with workouts. Apple calls it Audio Intelligence, and it essentially gives the Watch another way of understanding what is happening around you. The Series 12 and Ultra 4 can recognize important sounds, automatically identify music, recover the last few seconds of speech you failed to catch, and even create summaries of conversations you want to remember later.
Setting aside the obvious privacy concerns, this can be useful in many scenarios and is more than just another AI feature.
Live Rewind is the one I can’t wait to try
We’ve all had that moment where somebody says something, and your brain managed to process none of it. It could be due to the noisy environment around you, or perhaps someone mentioned an unfamiliar name. Maybe you’re listening while doing something else and suddenly realize you’ve missed the one important instruction buried in the conversation.
This is where Live Rewind comes in as your second set of ears. Double-press the Digital Crown and the Watch can show a text snippet covering the previous 15 seconds of speech. You can read what you missed, ask Siri about it, and even save the snippet in the Siri app if you actually need it later. Apple claims that activating Live Rewind doesn’t start some endless transcription session in the background. Rather, it retrieves only that previous 15-second window, and the text disappears unless you deliberately save it or ask Siri about it.
The feature doesn’t seem massive, but with the alternative being a recorder or taking notes, I can see it being handy. Apple has also built in a very deliberate social cue. Live Rewind plays an audible chime even when the Watch is on silent or connected to headphones. There’s also a full-screen animation, and a microphone indicator appears when you activate it. So anyone nearby gets an indication that you’ve asked the Watch to recover what was just said.
Audio Intelligence does much more than remember conversations
Siri Recap is naturally going to attract most of the attention because an Apple Watch ambiently taking notes about your conversations sounds futuristic and slightly alarming. But there’s a lot more to Audio Intelligence than just listening in on your conversation. Sound Recognition can detect things like alarms, sirens, doorbells, and a crying baby, then send an alert to the Watch even when the paired iPhone isn’t nearby. Apple primarily designed it as an accessibility feature for people who are deaf or hard of hearing, which is where its importance really lies.
You also have automatic Music Recognition with Shazam. Hear a song that caught your interest? You don’t need to open Shazam anymore, as you can simply have the new Apple Watches identify music playing around you automatically. The song and artist can then appear in the Music Recognition widget inside the Smart Stack without you tapping anything. It’s one of those tiny conveniences that just saves you a bit of time and effort, but I can already see how often it will get used.
Yes, Siri Recap is part of this too
We’ve already written at length about Siri Recap and why Apple’s privacy approach made us more comfortable with the idea of a smartwatch listening throughout parts of my day. But the short version is that Siri Recap can generate a title, summary, and key points from conversations rather than preserving a transcript. You can schedule when it operates based on time and location, manually toggle it from Control Center, edit or delete the resulting summaries, and unsaved Recaps automatically disappear after seven days.
Apple claims that Audio Intelligence doesn’t create or retain audio recordings. Raw audio is handled inside a hardware-isolated Secure Exclave on the S11 chip and is inaccessible to watchOS, apps, the user, or Apple. For Siri Recap, a condensed text representation eventually goes to Private Cloud Compute to create the summary.
This is an actual reason to buy the new hardware
Audio Intelligence can’t be simply installed on last year’s Watch. It requires the S11 chip and its Secure Exclave, limiting the complete feature set to the Apple Watch Series 12 and Ultra 4. An Ultra 3 can run watchOS 27 and gets Siri AI, but it doesn’t suddenly gain the other new features like Live Rewind, Siri Recap, Sound Recognition through Audio Intelligence, or the new automatic Shazam functionality.
Live Rewind and Siri Recap arrive in beta later in 2026, starting in English, and won’t initially be available in the EU. Apple also says some server-assisted Audio Intelligence functionality is subject to daily usage limits. It is also worth noting that Live Rewind and Recap need a compatible Apple Intelligence-enabled iPhone alongside the new Watch. So I’m not declaring any of these features brilliant before I’ve actually used them.
