VQ based comparative analysis of MFCC and BFCC speaker recognition system

摘要

The language is considered the important mode of communication in this world and key mechanism manipulated for the language is speech. The parametric form of signal is used by speech recognizers to get the key characteristics of speech signal for the purpose of recognition. Development in the digital signal based speech processing technology open the door of the design of highly potential based speaker recognition system. A speaker-recognition scheme is more flexibly capable to function starved of both, the support of clear user and independency of the oral noise. In this paper, the performance of Mel Frequency Cepstral Coefficient (MFCC) and Bark frequency Cepstral coefficient (BFCC) were analyzed their effects in a text dependent speaker recognition system using VQ vector quantization method. It is found that the MFCC is offer better recognition rate as contrasted to BFCC using VQ vector quantization as speaker modeling technique.