Imagine you're at a noisy party, straining to hear your friend's story over the din of chatter and music. What if you could simply look at them for a few seconds, and suddenly their voice was crystal clear while everything else faded into the background?
That futuristic scenario is now one step closer to reality, thanks to a team of researchers at the University of Washington. They've developed an AI-powered headphone system that can "enroll" a specific speaker with just a brief glance, then filter out all other sounds to focus solely on that person's voice.
Here's how it works: The wearer taps a button while looking directly at the person they want to listen to for 3-5 seconds. Microphones on the headset capture that speaker's unique vocal patterns and feed the data into an onboard machine learning system. From that point on, the AI latches onto the enrolled voice and plays only that audio back to the listener in real-time, even as they move around a noisy environment.
In tests with 21 subjects, the "Target Sound Hearing" system significantly outperformed regular headphones, with users rating the voice clarity nearly twice as high. While still just a prototype, the researchers say their invention could eventually make its way into commercial earbuds and hearing aids.
Previously:
• Audio while swimming is getting pretty good
