Reduce The Noise in Speech Signals Using Wavelet Filtering

Abstract

The capacity of the data channels is often reduced due to noise and distortion of the transmitted signals. Noise reduction is used in various areas (where from noise and distortion the transmitted signals cannot be isolated): speech / speaker recognition, image processing, mobile communication systems, medical signal processing, radio and radar systems, etc. This paper illustrates the problem of the presence of noise in speech signals. A model of additive white Gaussian noise is considered and adding it to the speech signal – modeling of noise process. The main features of wavelets, which used in noise reduction, are described. The main algorithm of the noise cancellation process using wavelet analysis techniques is considered. Carried out the practical implementation of noise reduction. The graphs of the original, noisy and cleaned signals are plotted. An analysis of the results of noise cancellation was carried out using different families of wavelets, graphs of the cross correlation of noisy and clean speech signals are plotted. Noise reduction carried out using Matlab programing.