Delta-MFCC Based Text-Independent Speaker Recognition System

  Shivangi Chaudhary,  Deepali Jain

Speaker Recognition is a technique that uses the acoustic features of the speech of the individual for his/her identification among the group of known/unknown speakers. Recognition of the particular individual is based on the information present in his/her voice. In this paper, MFCC (Mel Frequency Cepstral Coefficients) features and Delta MFCC features are extracted from the speech samples of the speaker. Paper includes comparative studies of the speaker identification results by Delta-MFCC features and MFCC features using Gaussian Mixture Model. In the present work, the advantage of using Delta-MFCC features over MFCC is proposed, as the MFCC only gives the power spectral envelope of the particular frame, to include the trajectories of MFCC coefficients over time delta features has been calculated.

Keywords- MFCC, Delta-MFCC, GMM

Unique Identification Number - IJEDR1403017
Page Number(s) - 2991-2993
Pubished in - Volume 2 | Issue 3 | Sept 2014
DOI (Digital Object Identifier) -   
Publisher - IJEDR (ISSN - 2321-9939)

  Shivangi Chaudhary,  Deepali Jain,   "Delta-MFCC Based Text-Independent Speaker Recognition System", International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.2, Issue 3, pp.2991-2993, Sept 2014, Available at :http://www.ijedr.org/papers/IJEDR1403017.pdf