Title :
Improved GMM-UBM/SVM For Speaker Verification
Author :
Liu, Minghui ; Dai, Beiqian ; Xie, Yanlu ; Yao, Zhiqiang
Author_Institution :
MOE-Microsoft Key Lab. of Multimedia Comput. & Commun., Univ. of Sci. & Technol. of China, Hefei
Abstract :
This paper combines Gaussian mixture model-universal background model (GMM-UBM) and support vector machine (SVM) through post processing the GMM-UBM scores of different dimension feature parameter with SVM in speaker verification. Because different dimension feature makes different contribution to recognition performance and SVM has good discriminability, this combining approach yields significant performance improvements on decision-making. Experiments on text-independent speaker verification in NISTO5 8conv4w-1conv4w data showed that the actual detection cost function (DCF) of the test system was reduced to 0.0290 from 0.0343
Keywords :
Gaussian processes; speaker recognition; support vector machines; GMM-UBM; Gaussian mixture model; SVM; detection cost function; speaker verification; support vector machine; universal background model; Cost function; Decision making; Laboratories; Multimedia computing; Pattern classification; Scalability; Speech; Support vector machine classification; Support vector machines; System testing;
Conference_Titel :
Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on
Conference_Location :
Toulouse
Print_ISBN :
1-4244-0469-X
DOI :
10.1109/ICASSP.2006.1660173
