Wavelet analysis is an effective tool for studying the local properties of signals for non-stationary signals with rapidly changing local frequencies. In our work an algorithms for numerical calculation of direct continuous fast WT with an arbitrary choice of scaling coefficients and algorithm for numerical calculation of the inverse fast WT and several more for WT calculations are developed for better signal analysis. The special approach is an analysis of automatic speech command recognition systems.