Wavelet Hybrid Threshold and Window Functions for Speech Enhancement

Following the introduction, Chapter-2 outlines the review of the most common and recently reported Speech enhancement techniques, with particular focus on Spectral Subtraction, Wiener filter and DWT Hybrid Thresholding the multitaper spectrum based Speech Enhancement algorithms which are the basis of this work. The shape of the analysis Window is also investigated in the present work, different adjustable Windows, the mathematical background and their important characteristics have been discussed in this chapter. This chapter also gives a detailed overview of standard six objective Speech quality measures that have been used in this work to quantify the results of the present work. In Chapter -3, the details of Spectral Subtraction algorithm and its drawbacks have been presented. To overcome the drawbacks Multiband Spectral Subtraction algorithm has also been discussed in this chapter. Comparative performance analysis has been exploited for Four Window viz., Kaiser Window, Dolph Chebyshev Window, Gaussian Window and recently proposed Polynomial Window used in both Spectral Subtraction and Multi Band Spectral Subtraction Speech Enhancement processes with different Practical noise