|
Our research has been focused on acoustic processing technologies to distinguish and to extract clean speech, building robust acoustic models that would endure unclear/spontaneous speech, developing widely-covered dictionaries and language models, all to create robust speech recognition systems that would work under any circumstances. Realizing the recognition of speech, speakers, and acoustic events, and to combine them with translation technologies would allow instant access to the events that take place around the world. "Speech" is a natural way of communication between humans to mutually understand each other; however, the language barriers prevent us from doing so. Ongoing research and development on translations from and between Japanese and multiple languages had taken place to overcome these barriers, and in 2010, along with organizations from around the world, the Universal Speech Translation Advanced Research Consortium, U-STAR ( http://www.ustar-consortium.com/ ) was founded to cover a wider range of languages. We have been promoting to enhance the speech translation technologies worldwide by connecting the speech translation servers around the world via network and to operate further research and development towards multi-lingual speech translation technologies by collecting real data from field experiments. Our research has also been focused on overcoming the communication barriers between the handicapped and the non-handicapped. We aim to support the handicapped and the elderly by utilizing speech communication technologies; such as transcribing speech into text and/or converting text into speech.
|
