Speaker Verification System Using LLR-Based Multiple Kernel Learning

  • Yi-Hsiang Chao
Conference paper
Part of the Lecture Notes in Electrical Engineering book series (LNEE, volume 240)


Support Vector Machine (SVM) has been shown powerful in pattern recognition problems. SVM-based speaker verification has also been developed to use the concept of sequence kernel that is able to deal with variable-length patterns such as speech. In this paper, we propose a new kernel function, named the Log-Likelihood Ratio (LLR)-based composite sequence kernel. This kernel not only can be jointly optimized with the SVM training via the Multiple Kernel Learning (MKL) algorithm, but also can calculate the speech utterances in the kernel function intuitively by embedding an LLR in the sequence kernel. Our experimental results show that the proposed method outperforms the conventional speaker verification approaches.


Log-Likelihood ratio Speaker verification Support vector machine Multiple kernel learning Sequence kernel 



This work was funded by the National Science Council, Taiwan, under Grant: NSC101-2221-E-231-026.


  1. 1.
    Reynolds DA, Quatieri TF, Dunn RB (2000) Speaker verification using adapted Gaussian mixture models. Digit Signal Proc 10:19–41Google Scholar
  2. 2.
    Rosenberg AE, Delong J, Lee CH, Juang BH, Soong FK (1992) The use of Cohort Normalized scores for speaker verification. Proc, ICSLPGoogle Scholar
  3. 3.
    Chao YH, Tsai WH, Wang HM, Chang RC (2006) A kernel-based discrimination framework for solving hypothesis testing problems with application to speaker verification. Proceedings of the ICPRGoogle Scholar
  4. 4.
    Auckenthaler R, Carey M, Lloyd-Thomas H (2000) Score normalization for text-independent speaker verification system. Digit Signal Proc. 10:42–54Google Scholar
  5. 5.
    Bengio S, Mariéthoz J (2001) Learning the decision function for speaker verification. Proceedings of the ICASSPGoogle Scholar
  6. 6.
    Wan V, Renals S (2005) Speaker verification using sequence discriminant support vector machines. IEEE Trans Speech Audio Proc 13:203–210Google Scholar
  7. 7.
    Campbell WM, Sturim DE, Reynolds DA (2006) Support vector machine using GMM supervectors for speaker verification. IEEE Signal Proc Lett 13Google Scholar
  8. 8.
    Karam ZN, Campbell WM (2008) A multi-class MLLR Kernel for SVM speaker recognition. Proceedings of the ICASSPGoogle Scholar
  9. 9.
    Rakotomamonjy A, Bach F.R, Canu S, Grandvalet Y (2008) SimpleMKL. J. Mach Learn Res 9:2491–2521Google Scholar
  10. 10.
    Herbrich R (2002) Learning Kernel classifiers: theory and algorithms, MIT PressGoogle Scholar
  11. 11.
    Chao YH, Tsai WH, Wang HM (2010) Speaker verification using support vector machine with LLR-based sequence Kernels. Proceedings of the ISCSLPGoogle Scholar
  12. 12.
    Luettin J, Maître G (1998) Evaluation protocol for the extended M2VTS database (XM2VTSDB). IDIAP-COM 98-05, IDIAPGoogle Scholar
  13. 13.
    Huang X, Acero A, Hon HW (2001) Spoken language processing. Prentics HallGoogle Scholar
  14. 14.
    Bengio S, Mariéthoz J (2004) The expected performance curve: a new assessment measure for person authentication. Proceedings ODYSSEYGoogle Scholar

Copyright information

© Springer Science+Business Media Dordrecht(Outside the USA) 2013

Authors and Affiliations

  1. 1.Department of Applied GeomaticsChien Hsin University of Science and TechnologyTaoyuanTaiwan

Personalised recommendations