Analysis of Intelligible Speech in a Noisy Environment

Currently, there exist hearing aids that distinguish speech from particular background noises, e.g., air-conditioning, but considerable work is required to enhance speech in a general noisy environment. In conjunction with DEKA Research and Development, our task is to create a test bed to evaluate such devices. The project consists of two parts: one is modeling conversation in a noisy environment, and the other is developing a measure for speech transmission quality. By obtaining and testing various sound data, we will determine the tolerance level for processing errors so that the signals sent to the user remain intelligible. We make extensive use of Fourier analysis in our mathematical study of acoustics, and work in the MATLAB technical computing environment.