Speech Tech Research Center to Open at Johns Hopkins
BALTIMORE -- The Johns Hopkins University here has been awarded a long-term, multimillion-dollar contract to establish and operate a new research facility that will primarily work toward new strategies related to speech technologies like speech recognition, speaker identification, speech analytics, and language processing.
The Human Language Technology Center of Excellence, which will receive a minimum of $48.4 million in funding from the U.S. Department of Defense through 2015, will be staffed by engineers, mathematicians, computer scientists, linguists, cognitive experts, and leading researchers from Johns Hopkins, the University of Maryland in College Park, and BBN Technologies, based in Cambridge, Mass.
Gary Strong, a former program manager and high-level official focusing on language technology projects at the National Science Foundation, will serve as the center’s executive director. James Baker, the co-founder and former CEO and chairman of Dragon Systems Inc., which in 1997 introduced the Dragon NaturallySpeaking dictation system (now owned by Nuance Communications), will serve as the center’s research director.
The center’s research will focus on advanced technology for automatically analyzing a wide range of speech, text, and image data in multiple languages. Strong has identified speech-enabled annotation as one of the first areas of concentration for the center, noting that right now such technologies require a lot of human interaction. “We should have automated and unsupervised annotation. It’ll be a challenge to get machines to do it alone, but we have reason to believe there is hope,” he says.
He also plans on looking at ways to make speech technologies more flexible and adaptable over time. “We’ll take an academic approach, taking products, understanding their architectures, and building platforms that will make them more flexible for the long term,” he says. “It’s about looking at developing new applications to do what current technologies cannot.”
Such work will have a strong impact on the federal government, which Strong says is “facing massive information overload. We need a better way to sort, filter, interpret, and call attention to important material that’s buried within the enormous amount of multilingual data being produced every day in other nations. The government does not have nearly enough people with the multiple language skills needed to review this material. We need to develop technology to help.”
Although the center will be addressing government needs, software developed by the center could have many commercial applications.