A new method for speech denoising and speaker verification using subband architecture

The purpose of this work is to estimate the cepstral coding on an arabic subbands system verification using dynamic time warping (DTW) method. The general principle is to split the whole frequency domain into several subbands on which statistical recognizers are independently applied. We propose two methods of spectral division: logarithmic decomposition and decomposition with filter-bank of two bands in arborescent structure. These two methods are very attractive because they are inspired by the functioning of the human hearing system. The strategy of decision rests on the individual decisions of recognizers in each subband. We also study which information is really used to recognize speakers. A comparison between the various variants of analysis are made to observe their performances.