Applications of multirate signal processing techniques in echo cancellation and noise suppression

Includes bibliographical references (pages [75]-77).This thesis studies the development of subband adaptive echo cancellation systems as well as noise suppression using wavelet transform. The theory and design issues of subbands and wavelets are explored and their application to echo and noise cancellation has been validated. Computer simulations have been carried out using estimated echo path, recorded far-end and near-end signals, and wavelet transform coefficients. A multiple channel subband adaptive echo canceler has been developed to deal with double talk detection in a hands-free cellular phone. A discrete cosine transform implementation of cosine-modulated quadrature mirror filter (QMF) banks has also been used and its performance compared with the traditional implementation of cosine-modulated QMF banks. Finally, wavelet transform has been used in denoising noise-corrupted speech in hearing aid design and its effectiveness was verified through simulations.M.S. (Master of Science