Newshot
Account

Apple details how iPhone 18 Pro / Max handles 'Hey Siri' to dispel always-listening concerns

AI summary

Apple published a new blog post detailing its new Secure Exclaves architecture for devices including the iPhone 18 Pro, and explaining the process behind 'Hey Siri.' On the iPhone 18 Pro and iPhone 18 Pro Max, Apple designed a flow for the 'Hey Siri' wake word in which a low-power first pass, an enclave re-check, and on-screen indicator verification must all occur before audio is released, reducing false wakes and letting users see when the device actually starts listening.

Translation(model translation, not the source's own text)

Breaking the always-listening doubts: Apple details the process behind Hey Siri on iPhone 18 Pro / Max

IT Home, November 12 news: Apple today (November 12) published a new blog post detailing the new "Secure Exclaves" architecture for devices such as the iPhone 18 Pro, and explaining in detail the process behind "Hey Siri."

According to the official blog post, on the iPhone 18 Pro and iPhone 18 Pro Max, Apple designed a process for the "Hey Siri" voice wake word of "low-power initial screening first, then review in the isolated area, and finally releasing audio only after verification through the on-screen indicator," thereby reducing false wake-ups and letting users intuitively see when the device actually begins listening.

Regarding the operational flow, IT Home cites the blog post and includes the following process:

  • Always-on low-power listening: The Always-On Processor (AOP) continuously listens to ambient sound, looking for speech segments that might be "Hey Siri." It only performs a rough first-round judgment and does not directly hand the complete recording to the main system.
  • Wake the main chip only on suspected wake: Once the AOP determines that "the wake word may have been heard," the iPhone 18 Pro and iPhone 18 Pro Max wake the main processor and hand the relevant audio buffer to the audio exclave, which avoids long-term high-power operation of the main chip.
  • The audio exclave performs echo cancellation first: The audio exclave performs processing such as echo cancellation on the audio to make the speech clearer, then hands the processed audio to the Siri exclave.
  • A second, more refined review: The Siri exclave runs a larger, more refined wake-word recognition model to further confirm whether the user really said "Hey Siri." Apple says this step filters out far more false positives than the AOP's judgment alone.
  • Siri is "woken" only after confirmation: If the wake word is confirmed, the Siri exclave sends only a "wake" flag to the Siri daemon in the main system; at this point no audio is transmitted to it.
  • Request the microphone and display the indicator dot: The Siri daemon then requests microphone permission from the audio driver, and the audio driver calls the interface to draw the microphone indicator dot and applies to the "Exclave Indicator Controller" (EIC) for audio access.
  • The on-screen indicator must pass verification: The EIC requires the secure display exclave to check the microphone indicator dot rendered on the screen, including its color, shape, display duration, minimum brightness, and whether nearby pixels obscure it. Only after verification passes does the system switch from "ambient mode" to "active mode" and allow audio to flow out.
  • Only then is audio handed to Siri: After verification passes, the Siri exclave hands the historical audio containing "Hey Siri" and the subsequent request content to the Siri daemon. The indicator dot remains displayed until the Siri session ends, for a minimum of 3 seconds.

Does the iPhone 18 Pro and iPhone 18 Pro Max record audio and video in the background all the time in order to respond to "Hey Siri" at any moment? In response, Apple said in the blog post that the background uses a burn-after-reading processing mechanism: under the always-on low-power listening state, the raw images and sounds captured by the camera and microphone are sent directly into the completely independent Secure Exclaves isolated area on the chip.

After data enters the isolated area, AI is only responsible for simplifying complex audio and video into minimal signals (such as "someone is coming," "this is family member A," "the wake word was heard"). Once recognition is complete, the raw audio and video data is immediately destroyed.

Apple also promises that "no one can obtain the raw data": not only can third-party apps not obtain it, but even the iOS / macOS main system, and even Apple itself, cannot access or extract the raw audio and video streams in the isolated area.

  • Secure Exclaves and Ambient Sensing

Advertising statement: External redirect links contained in the article (including but not limited to hyperlinks, QR codes, passwords, and other forms) are used to convey more information and save selection time; the results are for reference only. All articles on IT Home that contain external links include this statement.

Related articles

  • CNET releases battery drain test of 32 phones: Apple iPhone 18 Pro Max has the best overall endurance
  • The Wall Street Journal estimates Apple iPhone 18 Pro storage upgrade premium at about 200%
  • Switch 2 exclusive game "Mario Kart World" has been demonstrated running on Apple iPhone 18 Pro
  • Weekly increase exceeds one million: latest sales data for Apple iPhone 18 Pro revealed, series cumulative total surpasses 4.08 million units
  • Apple iPhone 18 Pro running iOS 27.0.1 is "perfectly jailbroken": highest privileges can still be maintained after reboot
  • DPReview says Apple iPhone 18 Pro/Max manual shutter captures only a single frame, with increased noise in low-light environments

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by IT之家
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗