URBAN HEALTH NETWORK

Not Available Systran Federal Corporation (SFC), the newly spun-off sister company of Systran Corporation, a Products Development Company with worldwide marketing and sales, and Wright State University (WSU), Dayton, Ohio, with expertise in reconfigurable mapping of algorithms are proposing to develop a hand-held one-way voice communication system. We have proposed to solve the speech recognition and the speech translator problem. We have proposed two feasible approaches. In the first approach, we have proposed to use Hidden Markov Models and its enhanced versions for speech recognition. We have proposed to enhance the speech signal because of the presence of high levels of noise in the target situations. The speech recognition as well as the speech enhancement functions will be mapped to hardware and will reside in a FPGA board. We have proposed to use systolic arrays to implement the Hidden Markov Model based speech recognition function. In the second approach, an existing integrated circuit which is built using neural networks for speech recognition will be used. It also contains all the ancillary circuits that are required for speech processing. In Phase I, both approaches will be assessed for performance, accuracy and other relevant requirements.