Speaker recognition using Vector Quantization by MFCC and KMCG clustering algorithm

Abstract

Speaker identification system is one of the applications of biometric using voice signal. In this paper we have implemented a speaker recognition system using a combination of Mel Frequency Capestral Coefficients (MFCC) & Kekre's Median Codebook Generation Algorithm (KMCG). The MFCC algorithm is used for feature extraction while the KMCG algorithm plays important role in code book generation and feature matching. For implementation simplicity the system is built as a text dependent system, i.e. common text used by all users. KMCG algorithm provides implementation simplicity along with high level of accuracy.