• Posted May 28, 2024

AI Headphones Let Listeners Hear Just One Voice in a Crowd

New AI headphone technology can help people “tune in” to specific folks in a crowd, allowing them to better hear a speaker even in noisy environments.

With the system, called “Target Speech Hearing,” a user wearing headphones looks at a person speaking for three to five seconds to “enroll” them, researchers report.

The AI then cancels out all other sounds in the environment and just plays the enrolled speaker's voice, even if the listener is moving around and no longer faces the speaker.

“With our devices, you can now hear a single speaker clearly even if you are in a noisy environment with lots of other people talking,” said senior researcher Shyam Gollakota, a professor of computer science and engineering with the University of Washington.

The AI system mimics and improves the human brain's ability to focus on speech coming from a single person, researchers explained.

To use the system, a person wearing off-the-shelf headphones fitted with microphones taps a button while directly looking at someone talking.

The sound waves of the speaker's voice reach microphones on either side of the headset, and the signal is forwarded to AI installed in the headphones. 

The AI learns the speaker's vocal patterns and latches onto the person's voice, continuing to play it even if speaker and listener move around. The system's ability to focus on the speaker even improves as they keep talking, researchers said.

The research team tested the system on 21 people, who rated the clarity of the enrolled speakers voice nearly twice as high, on average, as their voice in unfiltered audio.

Researchers presented their findings recently in Honolulu at the ACM CHI Conference on Human Factors in Computing Systems. Findings presented at scientific meetings should be considered preliminary until published in a peer-reviewed journal.

The computer code for their proof-of-concept device is available for others to build on, but the system is not commercially available.

Currently, the system can only enroll one speaker at a time, and it can't enroll a speaker if there's another loud voice coming from the same direction, researchers said.

The team is working to expand the system to earbuds and hearing aids in the future.

More information

The Cleveland Clinic has more about hearing aids.

SOURCE: University of Washington, news release, May 23, 2024