This proposal will demonstrate the feasibility of exploiting core noise canceling technology of UmeVoice, Inc. in conjunction with research/development in windscreen design, dynamic adjustments to microphone sensitivity, and filtering to cancel residual signal noise in order to eliminate the effects of noise in chaotic aural environments on Automatic Speech Recognition (ASR) performance in mission critical situations. In collaboration with Vanderbilt University/ISIS, we propose building a modeling environment in which spatial auditory scenes are created that include a wide array of noise sources and (an) operator(s) equipped with a voice interface to the command/control system. This will provide formal evaluation of core noise canceling technology of UmeVoice, Inc. in a variety of complex auditory scenarios. We intend to develop a simulation environment based on the model that will allow us to answer complex design questions, including deriving the optimal design of the noise canceling ASR input device, defining optimal arrangement of multiple operators in a room, determining optimal locations for other audio sources, including intercom or auditory alarm systems in order to minimize their effects on the system. This research will provide an input signal free from environmental noise and unintentional speech for improving ASR use/performance
