Similarity Measurement for Speaker Identification Using Frequency of Vector Pairs

Abstract

Similarity measurement is an important part of speaker identification. This study has modified the similarity measurement technique performed in previous studies. Previous studies used the sum of the smallest distance between the input vectors and the codebook vectors of a particular speaker. In this study, the technique has been modified by selecting a particular speaker codebook which has the highest frequency of vector pairs. Vector pair in this case is the smallest distance between the input vector and the vector in the codebook. This study used Mel Frequency Cepstral Coefficient (MFCC) as feature extraction, Self Organizing Map (SOM) as codebook maker and Euclidean as a measure of distance. The experimental results showed that the similarity measuring techniques proposed can improve the accuracy of speaker identification. In the MFCC coefficients 13, 15 and 20 the average accuracy of identification respectively increased as much as 0.61%, 0.98% and 1.27%.