Enhancement of speech signal using diminished empirical mean curve decomposition-based adaptive Wiener filtering

摘要

During the last few decades, speech signal enhancement has been one of the wide-spreading research topics. Numerous algorithms are being proposed to enhance the perceptibility and the quality of speech signal. These algorithms are often formulated to recover the clear signal from the signals that are ruined by noise. Usually, short-time Fourier transform and wavelet transform are widely used to process the speech signal. This paper attempts to overcome the regular drawbacks of the speech enhancement algorithms. As the frequency domain has good noise-removing ability, the short-time Fourier domain is also aimed to enhance the speech. Additionally, this paper introduces a decomposition model, named diminished empirical mean curve decomposition, to adaptively tune the Wiener filtering process and to accomplish effective speech enhancement. The performances of the proposed method and the conventional methods are compared, and it is observed that the proposed method is superior to the conventional methods.