The effect of DC coefficient on mMFCC and mIMFCC for robust speaker recognition

摘要

In Speaker Recognition (SR) system, feature extraction is one of the crucial steps where the particular speaker related information is extracted. The state of the art algorithm for this purpose is Mel Frequency Cepstral Coefficient (MFCC), and its complementary feature, Inverted Mel Frequency Cepstral Coefficient (IMFCC). MFCC is based on mel scale and IMFCC is based on inverted mel (imel) scale. There are two another set of features we proposed as mMFCC and mIMFCC. In state-of-the-art system, we neglect the DC co-efficient of DCT from the feature set. In this paper, the DC coefficient and its effect on recognition accuracy on MFCC-IMFCC, as well as, mMFCC-mIMFCC has been studied. This has been verified on two standard different types of databases, like, YOHO for clean speech signal and POLYCOST for telephone based speech. The recognition accuracy of the proposed feature is better than their respective baseline feature when the DC coefficient was included, as well as, when it was not included.