Hasegawa-Johnson named ASA Fellow
Heather Punke, ECE ILLINOIS
- ECE Associate Professor Mark Hasegawa-Johnson has been named Fellow of the Acoustical Society of America.
- Hasegawa-Johnson focuses his research in the area of acoustic signal processing and speech production, including speech recognition.
- One current project is working on technology that could improve the lives of people living with disabilities like cerebral palsy that make communication difficult.
ECE Associate Professor Mark Hasegawa-Johnson has been named Fellow of the Acoustical Society of America. The award will be formally presented to him during the Society’s meeting in November.
“Fellows of a society made contributions over many years and have a fairly substantial career in that specialty,” explained Hasegawa-Johnson, a researcher in the Beckman Institute for Advanced Science and Technology and an affiliate of the Coordinated Science Lab. With this fellowship, Hasegawa-Johnson is recognized for his contributions to vocal tract and speech modeling.
Hasegawa-Johnson said he was happy to be named Fellow, and pointed out that the University of Illinois is a great place to do his research. “It’s certainly one of the best places in the world to do research that combines mathematics and science like I try to do with my research,” he said. “People [here] created a lot of the math that my students have been applying to speech and audio processing algorithms.”
For the past 20 years, Hasegawa-Johnson has been conducting research in the area of acoustic signal processing and speech production. A faculty member at Illinois since 1999, Hasegawa-Johnson has led a research team whose past work has involved improving machine learning and understanding speech and audio signals. Many of the machine learning algorithms, originally developed in the 1980s for two-choice classification problems, could not be applied to speech and audio because of the high computational cost. Hasegawa-Johnson worked to fix this problem by borrowing aspects from linguistics.
“There’s this theory of distinctive features in linguistics, which says that any given language chooses a particular set of primarily binary distinctions in order to organize [. . .] that language,” he explained. Essentially, the differences that separate one word from another in each language can be broken down to 1s and 0s that machines can understand. He used the theory to organize the learning problem, so that machine learning algorithms could be used without unreasonable complexity.
In fact, results of his research have found their way into speech recognizers running on cell phones and in call centers.
While his work has impacted industry already, Hasegawa-Johnson is currently working on technology that could improve the lives of people living with disabilities like cerebral palsy.
People with cerebral palsy have difficulty speaking because their muscles are clenched the majority of the time, which makes talking painful.
“It’s a very difficult problem to solve, to get a speech interface that works well enough to be useable for a person who’s speech is unintelligible [most of the time],” he said. “It’s beyond the amount of time that [large companies] want to expend on that problem.”
If he and his group solve this problem and develop a working technology, he has plans for the technology’s future. “This is an area where a product distributed by a University of Illinois startup company could have a tremendous impact, helping people who need the help,” Hasegawa-Johnson said.
Even with a Fellow title under his belt, Hasegawa-Johnson has no plans to stop doing what he loves: research. “It’s nice to work in an area where you can do things that are intellectually challenging and also have an effect on society and change people’s lives.”