CurveBezier / iStock
By Matthew Hutson
Somehow, even in a room full of loud conversations, our brains can focus on one voice in something called a cocktail party effect. But the harder you get – or the older you get – the harder it is to do. Now, researchers could have figured out how to solve this – with a machine learning technique called the cone of silence.
Computer scientists have trained a neural network, which roughly mimics brain wiring, to locate and separate the voices of several people speaking in a room. The network did this in part by measuring how long it took for the sounds to hit a group of microphones in the center of the camera.
When researchers tested their configuration with extremely loud background noise, they found that the cone of silence located two voices at 3.7 degrees from their sources, they reported this month at the online conference only on neural information processing systems. This compares to a sensitivity of only 11.5º for state-of-the-art technology. When the researchers trained their new system with additional voices, it did the same trick with eight voices – at a sensitivity of 6.3 ° – even though it never heard more than four at a time.
Such a system could one day be used in hearing aids, surveillance devices, speakers or laptops. The new technology, which can also track moving voices, could even ease your Zoom calls by separating and reducing background noise, from vacuum cleaners to stray children.