iOS 15 Sound Recognition Accessibility Features

Apple significantly expanded its accessibility suite in iOS 15 by upgrading Sound Recognition, an essential assistive tool designed for deaf and hard-of-hearing users. While previous versions could only identify a preset list of standard environmental noises, iOS 15 introduced the ability to recognize custom sounds, allowing users to train their devices to detect specific, non-standard alarms and household appliances.

Custom Alarm and Appliance Recognition

The headline addition to Sound Recognition in iOS 15 is the Custom Sounds feature. Users are no longer limited to Apple's pre-programmed sound catalog. Instead, the iPhone can now be trained to listen for unique acoustic signatures, specifically tailored to the user's environment:

  • Custom Alarms: Users can train the system to identify proprietary or unique electronic alarms, such as specific smoke detectors, carbon monoxide monitors, or digital timers.
  • Custom Household Appliances: The system can learn unique audio cues from everyday home devices, including the distinct chime of a washing machine, a dryer, or a microwave cycle completion.
  • Custom Doorbells: If a home uses a distinct or non-traditional chime, the device can record and register that specific tone rather than relying on standard buzzer sounds.

How the Feature Works

Training iOS 15 to recognize custom sounds takes place directly on the device using local machine learning.

  1. Open Settings and navigate to Accessibility.
  2. Select Sound Recognition and ensure the toggle is turned on.
  3. Tap Sounds, then choose Custom Alarm or Custom Appliance or Doorbell.
  4. Enter a name for the sound and tap Continue.
  5. Follow the on-screen prompts to record the sound multiple times so the device can accurately map the acoustic profile.

Once set up, the iPhone continuously listens in the background. When it detects a trained sound, it immediately sends a push notification, flashes the screen (or the LED flash, if configured), and triggers haptic feedback.

Privacy and Local Processing

To maintain user privacy, all audio processing for custom sound recognition occurs locally on the device using Apple’s Neural Engine. The microphone does not stream ambient audio to external servers or cloud services, ensuring that household sounds remain entirely private while the system actively monitors the environment.