Novel Earbuds Enhance Learning-Based Speech

University of Washington researchers are working to enhance speech recognition in real-time with their novel 'ClearBuds' end-to-end hardware and software system.

“ClearBuds differentiate themselves from other wireless earbuds in two key ways,” said co-lead author Maruchi Kim  , a doctoral student in the Paul G. Allen School of Computer Science & Engineering. “First, ClearBuds use a dual microphone array. Microphones in each earbud create two synchronized audio streams that provide information and allow us to spatially separate sounds coming from different directions with higher resolution. Second, the lightweight neural network further enhances the speaker’s voice.”

Transcript

00:00:00 hi i'm wearing a set of wireless earbuds called clear buds with more and more wireless earbuds like these we see many people taking calls on the go or in noisy environments for example you might be trying to record something and your roommate starts vacuuming right next to you in this system we stream our audio to a

00:00:17 phone where we run a lightweight real-time speech enhancement network let me turn that on for you now that's better not only can clearbots press background noises like these but it does a great job at suppressing other voices too this is because the left and right clear blood streams are time synced to the

00:00:39 phone compared to single channel models that only learn acoustic cues we have also spatial information to separate background voices as well hi yeah i'd like to order a uh so even if my roommate were to come sit next to me order a pizza we don't have to hear his dictionaries as well why not thank you for taking the time to hear

00:00:57 about clear buds if you want to learn more please read our paper