Speech Enhancement in Wavelet Domain using Principle Component Analysis and Enhancement Filters

The aim of speech enhancement is to improve the perceptual quality and intelligibility of the speech by reducing the background noise. This paper proposes a technique in wavelet domain to enhance the signal. The signal is decomposed into approximation coefficients and detail coefficients which are filtered separately using spectral subtraction and wiener filter. The signal is reconstructed by transforming it into time domain by applying inverse wavelet domain. Wavelet features of the noisy speech signal are extracted and the dimension of the features is reduced using Principle component analysis (PCA). Experiments are conducted on noisy speech signal database (NOIZEUS), which consists of speech signals corrupted by eight different real world noises recorded at different signal-to-noise (SNR) levels. The performance of the proposed algorithm is evaluated using SNR, which is a standard measure of the amount of background noise present in a speech signal and Mean opinion score(MOS) .Experiments results show the increase in the efficiency of the proposed enhancement algorithm.