One thing our brains do really well is pick out a single voice from background noise. Unfortunately, this is a task that many hearing aids struggle with; though advanced hearing aids can pick out voices from noise like street traffic, they have a much harder time when the background noise is more voices, like at Virginia Street Brewhouse. This is known as the cocktail party problem.
Researchers at Columbia University in New York City are setting out to fix this problem. In fact, they’ve developed artificially intelligent (AI) technology that can amplify the correct speaker in a group.
Initial Research
In 2017, the Columbia research team created a system that could pick out a desired voice to amplify over other voices, but there was one major limitation: the device had to be pretrained to recognize specific speakers. So if you were in a restaurant with your family, your spouse’s voice would be amplified, but as soon as the waiter came over to recite the day’s specials, you wouldn’t necessarily be able to switch your attention.
Funding from Columbia Technology Ventures allowed the team to improve their original algorithm.
New Research
In 2019, the Columbia research team published a study detailing their new and improved technology. This complex system utilizes speech-separation algorithms with neural networks, which are complex mathematical models that imitate the brain’s natural computational abilities. It also monitors the wearer’s brain waves so it can boost the voice they want to focus on.
The device works by separating the voices of individual speakers from a group, then comparing the voices of each speaker to the brain waves of the person listening. The speaker whose voice pattern most closely matches the listener’s brain waves is amplified.
“By creating a device that harnesses the power of the brain itself, we hope our work will lead to technological improvements that enable the hundreds of millions of hearing-impaired people worldwide to communicate just as easily as their friends and family do,” explained senior study author Nima Mesgarani, Ph.D.
Next Steps
The next steps for the research team are to refine the algorithm so it can function in a broader range of environments as well as transform their prototype into a noninvasive device that can be worn externally on the head or ears. For more information about today’s hearing technology or to schedule an appointment, call LeMay Hearing & Balance today.