Connect with us

Technology

Researchers Develop AI Noise-Canceling Headphones That Let You Control What You Hear

courtesy of bgr.com

Select What Noises You Want to Hear

Noise-canceling headphones have made significant advancements with their transparency modes, allowing users to hear real-world noise while listening to music. However, they lack the ability to control what you hear. Now, researchers have created a set of AI noise-canceling headphones that give you the power to select which noises you want to hear. This allows you to filter out unnecessary sounds while remaining aware of important things.

Semantic Hearing System

These new headphones feature a system called "semantic hearing" that streams captured audio to a connected smartphone. Through the smartphone app and voice commands, users can choose from 20 different classes of sounds to let through the filters. These classes include sirens, baby cries, speech, vacuum cleaners, and even bird chirps. Watch the embedded video below for more details on how the system works.

Hopes for Commercial Release

The researchers presented their findings for the AI noise-canceling headphones at the UIS '23 conference in San Francisco. They express their intention to release a commercialized version of the headphones in the future, making this technology available to everyday users.

Control Your Environment

This breakthrough in noise-canceling technology is crucial, particularly for those who use headphones in public settings. With the help of AI, these headphones allow you to have control over what you hear. You can still shut yourself off from the world, but also be aware of sounds that may indicate danger or other important noises like a baby crying.

Positive Test Results

The headphones were tested in various environments, and the semantic hearing system proved to be effective. Out of the 22 participants, the target sound was rated as high quality compared to the original recording. While the system occasionally struggled to distinguish between vocal music and human speech, the researchers believe that more real-world data will help improve its performance.

Continue Reading