Particle.news

Download on the App Store

AI Headphones Isolate a Single Voice in Noisy Environments

University of Washington researchers develop a system that enhances auditory focus using AI and off-the-shelf headphones.

A University of Washington team has developed an artificial intelligence system that lets a user wearing headphones look at a person speaking for three to five seconds and then hear just the enrolled speaker’s voice in real time even as the listener moves around in noisy places and no longer faces the speaker. (University of Washington Photo / Kiyomi Taguchi)
A University of Washington team has developed an artificial intelligence system that lets a user wearing headphones look at a person speaking for three to five seconds and then hear just the enrolled speaker’s voice in real time even as the listener moves around in noisy places and no longer faces the speaker. Pictured is a prototype of the headphone system: binaural microphones attached to off-the-shelf noise canceling headphones.
Headphones that use AI to single out one voice in a crowded space
Image

Overview

  • The system, called Target Speech Hearing, lets users focus on one voice by looking at the speaker for three to five seconds.
  • AI processes the speaker's vocal patterns and cancels out other noises, even if the listener moves around.
  • The technology aims to improve conversations in noisy settings and assist those with hearing challenges.
  • Initial tests show significant improvement in voice clarity, with users rating it nearly twice as high as unfiltered audio.
  • Researchers plan to expand the system to earbuds and hearing aids, making the code publicly available for further development.