Similarity analysis of voice signals using wavelets with dynamic time warping

In this paper several wavelets are used with Dynamic Time Warping and Derivative Dynamic Time Warping techniques for analysis of sample voice signals. Statistical methods are used to determine the similarity of voice signals in the wavelet domain. Experiments conducted in this research include analysis of similar voice signals spoken by the same speaker, analysis of similar voice signals spoken by different speakers, and the study to determine the effect of complexity of voice signals on the similarity analysis. For the purpose of this research four subjects, two females and two males, are selected to speak the sample words. The limited number of experiments conducted in this research provided important information on the effectiveness of different wavelets and Time Warping techniques in successful analysis of similar sound signals. Some of the results are presented in tables that show the correlation between different voice signals, subjects, and/or techniques used in our analysis.