Noise-cancelling headphones that alert you to danger with AI?

While using noise-canceling headphones can effectively suppress baby cries, it can also erase sounds you really need to hear, like the horn of a nearby car in heavy traffic.

To make noise-cancelling headphones safe, a team at Columbia University is experimenting with the ability to recognize sounds that signal an imminent danger and trigger warnings to wearers. A research team at Columbia University’s Data Science Research Institute has improved the noise-canceling headphones function, which is a standard over-ear type, and added an outward-facing microphone on both sides of the wearer’s head. The microphone detects noise, and the sound is sent to an app running on the connected smartphone for processing.

Machine learning is used here, and the app can recognize and differentiate the cacophony of sounds around the headphones, as well as assess potential threats by sound. The sound of music flowing and the sound for the visually impaired at the crosswalk are not a threat. But something like a growing motor noise suggests that a car is approaching and judges it as an obvious danger. When it detects this sound, it sends a beep to the headphones.

The research team is currently conducting headphone experiments on pedestrians in New York on high-risk roads. However, although it can detect the danger to some extent, it doesn’t matter. The research team works with professors of psychology to try to specify the beep sound at the most effective point in order to attract user attention and give sufficient time to avoid risk. Related information can be found here .