Speech recognition system using burg method and Discrete Wavelet Transform

This work presents an automatic recognition system of the English spoken digits using a parametric method of power spectrum density estimation. The signal is pre-processed by using the endpoint detection algorithm and then we apply the Discrete Wavelet Transform to extract the approximation coefficients. Burg algorithm is then applied to estimate the power spectrum density. At the stage of matching, the similarity measurements between signals are used to this purpose. The developed algorithm gives a good accuracy rate.