Speaker Verification Based on GMM Multidimensional Likelihoods and SVM
LIU MingHui, DAI BeiQian, XIE YanLu
MOEMicrosoft Key Laboratory of Multimedia Computing and Communication, University of Science and Technology of China, Hefei 230027 Department of Electronic Science and Technology, University of Science and Technology of China, Hefei 230026
Abstract In this paper, a textindependent speaker verification system based on GMM multidimensional likelihoods and SVM is proposed, which combines the advantages of both generative model and discriminative model. In this method, the GMM multidimensional likelihoods for the test speech are regarded as new features for SVM. Experiment results of textindependent speaker verification on NIST'05 8conv4w1conv4w database show effectiveness of the proposed system.
[1] Vapnik V N. The Nature of Statistical Learning Theory. Berlin, Germany: SpringerVerlag, 1995 [2] Burges C J C. A Tutorial on Support Vector Machines for Pattern Recognition. Data Mining and Knowledge Discovery, 1998, 2(2): 121167 [3] Huang Wei. Dai Beiqian. TextIndependent Speaker Verification Based on GMM Statistical Parameters and SVM. Journal of Data Acquisition and Processing, 2004, 19(4): 365370 (in Chinese) (黄 伟,戴蓓.基于GMM 统计特性参数和SVM 的话者确认.数据采集与处理, 2004, 19(4): 365370) [4] Wan V, Campbell W M. Support Vector Machines for Speaker Verification and Identification // Proc of the International Workshop of Neural Networks for Signal Processing. Sydney, Australia, 2000, Ⅹ: 775784 [5] Wan V, Renals S. SVMSVM: Support Vector Machine Speaker Verification Methodology // Proc of the IEEE International Conference on Acoustics, Speech and Signal Processing. Hong Kong, China, 2003, Ⅱ: 221224 [6] Bengio S, Mariethoz J. Learning the Decision Function for Speaker Verification // Proc of the IEEE International Conference on Acoustics, Speech and Signal Processing. Salt Lake City, USA, 2001, Ⅰ: 425428 [7] Liu Minghui, Dai Beiqian, Xie Yanlu, et al. Improved GMMUBM/SVM for Speaker Verification // Proc of the International Conference on Acoustics, Speech and Signal Processing. Toulouse, France, 2006, Ⅰ: 925928 [8] Fine S, Navratil J, Gopinath R A. Enhancing GMM Scores Using SVM “Hints” // Proc of the 7th European Conference on Speech Communication and Technology. Aalborg, Denmark, 2001: 17571760 [9] Reynolds D A. Speaker Identification and Verification Using Gaussian Mixture Speaker Models. Speech Communication, 1995, 17(1/2): 91108 [10] Doddington G R, Przybocki M A, Martin A F. The NIST Speaker Recognition Evaluation-Overview, Methodology, Systems, Results, Perspective. Speech Communication, 2000, 31(2): 225254 [11] Reynolds D A. Speaker Verification Using Adapted Gaussian Mixture Models. Digital Signal Processing, 2000, 10(1/2/3): 1941 [12] The NIST Speaker Recognition Evaluation [DB/OL]. [20061018]. http://www.nist.gov/speech/tests/spk/index.htm
2008, Vol. 21 
