Dr. Andrew Sears’ perspectives on voice recognition featured in UK news daily, The Independent

Date Posted: - 11/6/2009

Dr. Andrew Sears, Professor and Chair of The Department of Information Systems at UMBC was recently featured in the “Tech and Gadgets” section in the October, 25th edition of The Independent, a London-based news daily. The article, Flying cars and heated pants: Shouldn’t we be able to buy this stuff by now?, examines contemporary and emerging technologies and the products promised consumers over the years, and asks the proverbial question, “What’s the hold up?”

The Independent’s writer, Tim Walker, reports on the marketplace whereabouts of more seemingly sci-fi items such as vehicles that remove themselves from traffic jams by lifting off vertically and then flying away. However, his article also touches on the status of Voice-recognition software that works—something the public may feel should already be here. UMBC’s Dr. Andrew Sears helps clear up what appears to be working and identifies many of the challenges that have yet to be solved algorithmically in the segment republished below. The entire article may be viewed by visiting the following link: http://www.independent.co.uk/life-style/gadgets-and-tech/features/flying-cars-and-heated-pants-shouldnt-we-be-able-to-buy-this-stuff-by-now-1807362.html

When the speech-to-text service Spinvox was caught using call-centre workers to help transcribe voice messages into texts, rather than the promised computer algorithms, it left many wondering why something that should be so simple is such a challenge.

"Spoken language is very complex," says Professor Andrew Sears, Director of the Interactive Systems Research Centre at the University of Maryland, Baltimore County (UMBC). "There are complex grammar rules that may not be followed while we're speaking. There's evolving vocabulary. There's context-dependent speech, where you miss a word but are able to fill the gaps. Computers simply aren't able to do that yet."

Until such a time as they are, developers will stick to specialist programmes for narrower linguistic realms, such as medical or legal transcription. Sears' own work is focused on making communications technology easier for disabled people to use. Eventually, these specialist solutions will be integrated into more general-purpose software.

"There are key problems to be addressed," he adds. "There's still too much training required before these systems become sufficiently accurate for most people to be willing to use them. And we need software that can learn how it mishears certain words and recalibrate itself accordingly."

Dr. Sears’ research focus is in Human-Centered Computing (HCC). He is also Co-Editor-in-Chief of Transactions on Accessible Computing (TACCESS), a quarterly journal that publishes refereed articles addressing issues of computing as it impacts the lives of people with disabilities. Additional information about Dr. Sears’ work may be found at http://userpages.umbc.edu/~asears/.