Development of a Digital Auditory/Speech Processor

The objectives of phase one of this project are as follows: 1) to develop the digital signal processing algorithms necesary to improve the intelligibility of speech degraded by noise, 2) to develop the software that will implement these algorithms, and 3) to evaluate the effectiveness of the speech enhancement software on one human subject in different laboratory noise environments using a Motorola 56001-based workstation. The algorithms developed in this research effort will be based on the critical band model of the human ear. The critical band model has been the basis of research efforts in the past concerning noise-induced hearing loss, presbycusis, and speech recognition. Little effort, however, has been made to use critical band theory in the field of speech enhancement. The effectiveness of the speech enhancement system developed in this effort will be judged both objectively and subjectively. A battery of speech recognition tests will be employed in several noise environments under several signal-to-noise ratios. Subjective quality assessments from the listeners will also be obtained. Although the emphasis of this research will be on speech enhancement, the algorithms developed will be applicable to any known or predictable target signal.