Noise is omnipresent – be it the construction noise on the street, the noisy neighbors or the hustle and bustle in a café. Conventional noise cancellation systems in headphones can block out background noise, but they do not differentiate between relevant and irrelevant noises. A team of researchers at the University of Washington has now developed a novel solution that allows users to filter out only certain ambient noises.
Targeted noise selection thanks to artificial intelligence
The technology combines noise cancellation with a neural network running on a smartphone. This can recognize 20 different ambient noises such as alarm clocks ringing, babies, sirens or birds chirping. The user can then choose which sounds they want to hear – all others are filtered out.
“Imagine you are in a park enjoying the birds singing, but are disturbed by a noisy group. With our system, your headphones could let the birdsong through and block out the rest,” said Shyam Gollakota of the University of Washington in one Press release.
Real-time processing of filters
However, implementing this idea was a real challenge: “First, we needed a highly developed intelligence to identify all the different ambient noises,” the researcher continued. “Then we had to separate the desired sounds from all the distracting noises. Last but not least, the extracted sounds had to be output in real time and without delay to match the visual impressions.”
Using the algorithms developed, the researchers managed to process audio signals in less than a hundredth of a second. In addition to ambient noise detection, the technology can also be used to filter out voices from background noise and thus simplify conversations.
Gollakota sees its developments as the beginning of a new generation of intelligent hearing systems: “We have a unique opportunity to shape the future of intelligent hearing aids that can improve human hearing and make our lives easier.”
picture of Gigxels.com on Pixabay