DocumentCode :
3183810
Title :
Hybrid of wavelet and MFCC features for speaker verification
Author :
Kumar, Pawan ; Chandra, Mahesh
Author_Institution :
Dept. of ECE, Cambridge Inst. of Technol., Ranchi, India
fYear :
2011
fDate :
11-14 Dec. 2011
Firstpage :
1150
Lastpage :
1154
Abstract :
In this paper Wavelet Based Mel Frequency Cepstral Coefficient (WMFCC) features are proposed for speaker verification. The performance of WMFCC features is evaluated and compared with the performance of Mel Frequency Cepstral Coefficient (MFCC) features. A database of ten Hindi digits of sixteen speakers is used during simulation of results. Gaussian Mixture Models (GMMs) are used for maximum log likelihood calculation during verification. The proposed features have shown an increment of 1.18% in performance over MFCC features for text dependent speaker verification system.
Keywords :
Gaussian processes; maximum likelihood estimation; speaker recognition; wavelet transforms; GMM; Gaussian mixture model; WMFCC feature; maximum log likelihood calculation; mel frequency cepstral coefficient; speaker verification; wavelet feature; Feature extraction; Filter banks; Hidden Markov models; Mel frequency cepstral coefficient; Speech; Speech recognition; Wavelet transforms; GMM; Hindi digits; MFCC; WMFCC;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Information and Communication Technologies (WICT), 2011 World Congress on
Conference_Location :
Mumbai
Print_ISBN :
978-1-4673-0127-5
Type :
conf
DOI :
10.1109/WICT.2011.6141410
Filename :
6141410
Link To Document :
بازگشت