Transform based speech enhancement techniques

This thesis presents a detailed study on speech enhancement algorithms in the trans-form domain. The various effects of noise on speech are discussed. The main objectives of this work are to attenuate the noise component of a noisy speech in order to en-hance the quality of the processed speech using transform based filtering algorithms, and to carry out a comprehensive evaluation and comparison of their performances. A literature review of various speech enhancement techniques with a focus on those implemented in the transform domain is presented. The strengths and weaknesses of various transforms such as the Discrete Fourier Transform (DFT), Discrete Co-sine Transform (DCT), Karhunen Loeve Transform (KLT) and Wavelet Transform (WT) are discussed. It also presents a number of commonly used speech enhancement schemes, such as magnitude spectral subtraction, power spectral subtraction, Wiener filter and Minimum Mean Square Error filter.Doctor of Philosophy (EEE