A Wavelet Packet Speech Denoising using Spectral Entropy

M. Talbi, H. Belgacem, and A. Cherif (Tunisia)


Enhancement, spectral subtraction, SNR, discrete wavelet packet transform, spectral entropy, histogram


In this work, we are interested in developing of speech denoising tool by using a discrete wavelet packet transform (DWPT). For noise reduction, a part of wavelet packet coefficients are set to zero and the others are thresholded. For estimating the non stationary noise level we employ the spectral entropy. To evaluate approach, we make a comparison between our technique and spectral subtraction. The experimental implementation uses speech corrupted by two sorts of noise, white and Volvo noises.

Important Links:

Go Back