Graduate student Ed Chau, left, is working with Prof. Bob Dony to help develop a hearing aid that could screen out background noise.

PHOTO BY DEAN PALMER/
SCENARIO IMAGING

Helping to solve the biggest problem affecting people who wear hearing aids is more than academic for Ed Chau, a master's student in the School of Engineering. Among the millions of people who could benefit from more powerful hearing aids able to screen out distorting background noise is Chau himself.

  "I'm not happy with what I have," he says, referring to the devices he has worn for the past seven years to compensate for hereditary hearing loss.

  Bringing to Guelph an industrial scholarship worth $13,800 a year, he hopes to help a young Waterloo-based company develop the computing smarts for integrated circuits to be used in digital hearing aids, which promise greater acuity than conventional analogue devices.

  "Technically, he has excellent qualifications to do this research," says engineering professor Bob Dony of his new graduate student. "The fact that he wears hearing aids is an interesting addition to his complement of interests."

  Newly arrived this fall, Chau carried a post-graduate industrial scholarship jointly funded by the Natural Sciences and Engineering Research Council, U of G and dspfactory Ltd., a recent spinoff of hearing aid manufacturer Unitron in Kitchener. Established about 18 months ago, the company makes integrated circuits that perform extremely low-powered digital signal processing for potential use in a range of applications.

  About five million hearing aids are fitted every year worldwide, says dspfactory president Dan Murray. Beyond those devices, he says his company's technology might power computer chips used in cellular phones, portable speech recognition systems, speech recorders for portable personal organizers and noise-reducing headsets.

  Chau came to Guelph after discussions involving Dony and Todd Schneider, dspfactory's vice-president, technology (the two met several years ago in Waterloo).

  Under the terms of his scholarship, Chau spends one day a week working for dspfactory, following up on two terms he spent there during undergraduate co-op placements from the University of Waterloo.

  "I was looking specifically for hearing aid companies," he says. "Ideally, I'll be able to develop a superior neural network algorithm for use in audio signal processing."

  Most hearing aids still use analogue technology to amplify sounds, but Schneider says the digital market is expanding rapidly, much as an earlier wave overtook analogue telecommunications a decade ago. "The computing power we put in your ear is as much as in desktop PCs of a couple of years ago," he says.

  His company plans to take its prototype digital chip for hearing aids into full-scale production in about six months.

  That's music to the ears of people such as Chau, who says digital processing power is the key to solving the biggest affliction for users of hearing aids: background noise.

  "A lot of the problem that people have with hearing aids is that speech perception or speech intelligibility decreases dramatically from a quiet to a noisy environment."

  Processing audio signals properly requires millions of mathematical calculations, a formidable challenge for a miniature chip nestled inside a hearing aid or a cell phone.

  "There's so much computational power in such a small package," says Dony. "To get it so tiny, you're really pushing the limits of integrated circuit technology."

  At first glance, this project looks like a departure for Dony, who studies image processing for medical imaging, remote sensing and telerobotics as part of the Intelligent Systems Laboratory in the School of Engineering. He is also a member of the Guelph Natural Computation Research Group linking faculty in his school with the Department of Computing and Information Science.

  "I've done a lot of work with images," he says. "Those approaches can be applied to audio and speech signals. It's just a natural extension of the work I've been doing."

  Equipped with detectors capable of directional processing, tomorrow's hearing aids are expected to permit the user to focus on a conversation even in the middle of cocktail party rhubarb.

  "Whatever you face, the algorithm will amplify the sound from the front," says Chau, who had hearing aids fitted seven years ago after his impairment worsened. "If others around you aren't talking to you, you'll have less trouble because the hearing aid will help focus on sound coming from in front."

  His hearing loss is moderately severe, making him unable to pick up high frequencies in particular. "If the phone rings, I won't hear it. The lower frequencies are mostly OK, but I have some loss."

  Dony expects to use some of the U of G funding approved last year by the Canada Foundation for Innovation and Ontario Investment Trust to upgrade the School of Engineering's signal processing lab for this and related projects.