Modified mel-frequency cepstral coefficients (MMFCC) in robust text-dependent speaker identification

Abstract

In this paper a new approach in close-set speaker identification has been presented. The traditional melfrequency cepstral coefficient (MFCC) feature has been modified and it's name was given modified MFCC (MMFCC). Text-dependent dataset has been used to test the presented method's speaker identification rate. The speaker identification rate of the presented study was estimated in both clean and contaminated conditions. Four types of noises were added to clean signal to get noisy signals for a limit of signal to noise ratios (SNRs) from −5dB to 10 dB. Moreover, the obtained performance was compared with the performance of the traditional features like MFCC- and Gammatone frequency cepstral coefficient (GFCC)-based methods. The evaluated results showed the proposed method achieves significant improved performance over conventional MFCC- and GFCC-based methods performance under noisy conditions.