This article was written by DeLiang Wang.
My mother began to lose her hearing while I was away at college. I would return home to share what I’d learned, and she would lean in to hear. Soon it became difficult for her to hold a conversation if more than one person spoke at a time. Now, even with a hearing aid, she struggles to distinguish the sounds of each voice. When my family visits for dinner, she still pleads with us to speak in turn.
My mother’s hardship reflects a classic problem for hearing aid manufacturers. The human auditory system can naturally pick out a voice in a crowded room, but creating a hearing aid that mimics that ability has stumped signal processing specialists, artificial intelligence experts, and audiologists for decades. British cognitive scientist Colin Cherry first dubbed this the “cocktail party problem” in 1953.
More than six decades later, less than 25 percent of people who need a hearing aid actually use one. The greatest frustration among potential users is that a hearing aid cannot distinguish between, for example, a voice and the sound of a passing car if those sounds occur at the same time. The device cranks up the volume on both, creating an incoherent din.
It’s time we solve this problem. To produce a better experience for hearing aid wearers, my lab at Ohio State University, in Columbus, recently applied machine learning based on deep neural networks to the task of segregating sounds. We have tested multiple versions of a digital filter that not only amplifies sound but can also isolate speech from background noise and automatically adjust the volumes of each separately.
We believe this approach can ultimately restore a hearing-impaired person’s comprehension to match—or even exceed—that of someone with normal hearing. In fact, one of our early models boosted, from 10 to 90 percent, the ability of some subjects to understand spoken words obscured by noise. Because it’s not necessary for listeners to understand every word in a phrase to gather its meaning, this improvement frequently meant the difference between comprehending a sentence or not.
Without a better hearing aid, the world’s hearing will get worse. The World Health Organization estimates that 15 percent of adults, or roughly 766 million people, suffer from hearing loss. That number is rising as the population expands and the proportion of older adults becomes larger. And the potential market for an advanced hearing aid isn’t limited to people with hearing loss. Developers could use the technique to improve smartphone speech recognition. Employers could use it to help workers on noisy factory floors, and militaries could equip soldiers to hear one another through the noisy chaos of warfare.
It all adds up to a big potential market. The global US $6 billion hearing aid industry is expected to grow at 6 percent every year through 2020, according to the market research firm MarketsandMarkets, in Pune, India. Satisfying all those new customers, though, means finding a way to put the cocktail party problem behind us. At last, deep neural networks are pointing the way forward.
To read the rest of this article, click here.