A Vocal Tsunami

摘要

Voice recognition has become one of the leading forms of biometric identification, and it is important to ensure that this technique is as accurate as possible. The main algorithm for single word voice recognition is Dynamic Time Warping (DTW). In this study, we have approached the idea of DTW from a multiple input structure. By performing an analysis on n - 1 control samples and 1 test sample, we propose to take into account subtle variations in speech pattern to increase accuracy of testing. Through the application of the Tsunami Algorithm to an n-dimensional structure, we compute the shortest weighted path that will identify the relation between the samples. Our results show that comparison of paths with an n-dimensional diagonal give a more accurate method of single word verification than that of DTW.