Noise-canceling headphones are a lifesaver when it comes to living and working in noisy places. These headphones can identify background sounds and block them out, providing much-needed peace and quiet. However, traditional noise-canceling technology has a flaw – it can’t differentiate between unwanted background noise and important information, which means that headphone users can be oblivious to what’s happening around them. To address this issue, a team has developed a system for targeted speech hearing in noisy environments and created AI-based headphones that can selectively filter out specific sounds while still allowing others to come through. This innovation aims to provide headphone users with the ability to focus on the sounds they want to hear while blocking out the ones they don’t.
Traditional noise-canceling technology attempts to block out all sounds, including important information. Shyam Gollakota and his team at the University of Washington have developed AI-based headphones that can differentiate between different sounds in order to selectively filter them out. This technology allows for targeted speech hearing in noisy environments, providing a way for headphone users to be aware of their surroundings while still enjoying peace and quiet. Gollakota will be presenting this work on Thursday, May 16, at 1:20 p.m. EDT during a joint meeting of the Acoustic Society of America and the European Acoustics Association.The 177th Meeting of the Acoustical Society of America and the Canadian Acoustical Association will take place from May 13-17 at the Shaw Centre in downtown Ottawa, Ontario, Canada.
Gollakota explained, “Picture yourself in a park, enjoying the peaceful sounds of singing birds, but then you hear the loud chatter of a nearby group of people who won’t stop talking. Now, imagine if your headphones could help you focus on the sounds of the birds while ignoring the rest of the noise. That is exactly what we aimed to accomplish with our system.”
Gollakota and his team integrated noise-canceling technology with a smartphone-based neural network that was trained to identify and isolate specific sounds.In order to achieve our goal, we had to first develop a sophisticated system to recognize a wide range of environmental sounds,” explained Gollakota. “Then, we had to find a way to isolate the desired sounds from the background noise. And if that wasn’t challenging enough, we then had to filter out any unwanted sounds and play only the selected categories through the headphones in real time.” This seamless integration of the system was no simple feat.The information needed to be synchronized with the user’s sense of sight, as they cannot afford to be hearing someone speaking two seconds later. This requires the neural network algorithms to process sounds in real-time, within a fraction of a second, which is what we have accomplished.” The team utilized an AI-powered approach to specifically focus on human speech. Using similar content-aware techniques, their algorithm is able to recognize a speaker and separate their voice from background noise in real-time, resulting in clearer conversations. Gollakota is enthusiastic about being at the forefront of the next generation of audio devices. “We have a unique opportunity to shape the future of audio technology.Gollakota stated, “The development of smart hearables has the potential to improve human hearing and increase cognitive abilities, ultimately leading to a better quality of life.”
