In connection: Most new headphones have some form of noise cancellation. How it works is random. Apple’s AirPods are pretty good. Cheaper brands like Earfun are mediocre. But none seem to block external noise 100%.
Engineers at the University of Washington have developed a headphone that uses machine learning to achieve near-total noise cancellation. The headphones, dubbed ClearBuds, were recently unveiled at the Association for Computing Machinery’s International Conference on Mobile Systems, Applications and Services. Aside from the obvious application in audio wearables, the AI canceling technique could be used in home speakers and help robots track their location.
A short video (below) shows the headphones silencing a vacuum cleaner and even another person’s voice. The method effectively isolates the speaker’s voice without any audio interference. Other tested methods still let some background noise through. Of course, a practical demo would be more convincing.
Like other noise-cancelling technologies, ClearBuds use dual microphones to pick up speaker and external noise. However, the way it processes signals is completely different.
Maruchi Kim, a graduate student in UoW’s Paul G. Allen School of Computer Science and Engineering, explains that each earbud produces two synchronized high-resolution audio streams that contain data about the direction of each recorded sound. This technique allows the AI to create a spatial audio profile of the environment and isolate the speaker’s voice and noise sources more accurately than bi-directional microphones.
“Because the speaker’s voice is close and roughly equidistant from both headphones, the neural network was trained to focus only on their speech and to filter out background noise, including other voices,” the study co-author said. explained Ishaan Chatterjee. “This method is very similar to how your ears work. You use the time difference between sounds in your left and right ear to determine the direction the sound is coming from.”
Most quality earbuds have mics on each earbud, but Allen says only one is actively sending audio for processing at a time. With ClearBuds, each earbud continuously sends simultaneous audio streams. For this method, scientists had to develop a special Bluetooth network protocol for headphones that synchronizes the two streams within 70 microseconds of each other.
Although the ClearBuds are slightly larger than some of the most popular compact headphones available, the AI processing still needs to be done by a connected device that can run the AI. The team is working to make the neural network algorithms more efficient so the processing can happen on the headphones.
The researchers did not name a marketing plan. However, once their work is fully completed, there is a strong possibility that a commercial product will be manufactured or the technology licensed.