Determinación de la relación señal a ruido de la vo z utilizando la transformada de wavelet

The main purpose of this research was to develop a nd implement the algorithms to calculate the signal -tonoise ratio of a voice signal using the Wavelet Tra nsform. The Wavelet Transform of Daubechies (DB), Symlett (Sym) and Coiflet (Coif) with 3,5,10 and 15 levels of decomposition were applied. In order to test the developed algorithms, a signal with a known signal-to-noise r atio was used. Voice signals were also recorded and denoising was made using the mentioned transform. It was veri fied that the maximum denoising of the original sig nal was accomplished with 10 and 15 levels of decomposition, resulting in a processed signal without distorsio n and no background noise. The lowest signal-to-noise ratio values during the unvoiced segments of the recorded sentence were found using these values.