Researchers at the University of Washington have leveraged deep-learning algorithms, introducing a system dubbed "semantic hearing," enabling users to selectively filter sounds in real time.
Unlike conventional noise-canceling headphones that indiscriminately block out ambient noise, this approach allows users to choose which sounds they want to hear.
How the new AI noise-canceling technology works
The technology streams captured audio from the headphones to a connected smartphone, which then eliminates all environmental sounds. Through either voice commands or a dedicated smartphone app, users can specify the sounds they wish to include from a selection of 20 classes, such as sirens, baby cries, speech, vacuum cleaners, and bird chirps. Only the chosen sounds are then played through the headphones.
Presented at UIST '23 in San Francisco on November 1, the research team plans to release a commercial version of the system in the future.
When being tested across diverse environments (offices, streets, and parks) the semantic hearing system successfully extracted target sounds, such as sirens, bird chirps, and alarms, while eliminating other background noise.
