Advertisement

Data-Driven Impostor Selection for T-Norm Score Normalisation and the Background Dataset in SVM-Based Speaker Verification

  • Mitchell McLaren
  • Robbie Vogt
  • Brendan Baker
  • Sridha Sridharan
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5558)

Abstract

A data-driven background dataset refinement technique was recently proposed for SVM based speaker verification. This method selects a refined SVM background dataset from a set of candidate impostor examples after individually ranking examples by their relevance. This paper extends this technique to the refinement of the T-norm dataset for SVM-based speaker verification. The independent refinement of the background and T-norm datasets provides a means of investigating the sensitivity of SVM-based speaker verification performance to the selection of each of these datasets. Using refined datasets provided improvements of 13% in min. DCF and 9% in EER over the full set of impostor examples on the 2006 SRE corpus with the majority of these gains due to refinement of the T-norm dataset. Similar trends were observed for the unseen data of the NIST 2008 SRE.

Keywords

Support Vector Machine Support Vector Full Dataset Speaker Recognition Speaker Verification 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

References

  1. 1.
    Kajarekar, S.S., Stolcke, A.: NAP and WCCN: Comparison of approaches using MLLR-SVM speaker verification system. In: Proc. IEEE ICASSP, pp. 249–252 (2007)Google Scholar
  2. 2.
    Stolcke, A., Kajarekar, S.S., Ferrer, L., Shriberg, E.: Speaker recognition with session variability normalization based on MLLR adaptation transforms. IEEE Trans. on Audio, Speech, and Language Processing 15, 1987–1998 (2007)Google Scholar
  3. 3.
    Auckenthaler, R., Carey, M., Lloyd-Thomas, H.: Score normalization for text-independent speaker verification systems. Digital Signal Processing 10(1), 42–54 (2000)Google Scholar
  4. 4.
    Burges, C.: A tutorial on support vector machines for pattern recognition. Data Mining and Knowledge Discovery 2(2), 121–167 (1998)Google Scholar
  5. 5.
    McLaren, M., Baker, B., Vogt, R., Sridharan, S.: Improved SVM speaker verification through data-driven background dataset selection. To be presented in Proc. IEEE ICASSP (2009)Google Scholar
  6. 6.
    Nation Institute of Standards and Technology: NIST speech group website (2006), http://www.nist.gov/speech
  7. 7.
    Campbell, W., Reynolds, D., Campbell, J.: Fusing discriminative and generative methods for speaker recognition: Experiments on switchboard and NFI/TNO field data. In: Proc. Odyssey, pp. 41–44 (2004)Google Scholar
  8. 8.
    Campbell, W., Sturim, D., Reynolds, D., Solomonoff, A.: SVM based speaker verification using a GMM supervector kernel and NAP variability compensation. In: Proc. IEEE ICASSP, pp. 97–100 (2006)Google Scholar
  9. 9.
    McLaren, M., Vogt, R., Baker, B., Sridharan, S.: A comparison of session variability compensation techniques for SVM-based speaker recognition. In: Proc. Interspeech, pp. 790–793 (2007)Google Scholar
  10. 10.
    Bonastre, J., Wils, F., Meignier, S.: ALIZE, a free toolkit for speaker recognition. In: Proc. IEEE ICASSP, pp. 737–740 (2005)Google Scholar
  11. 11.
    Chang, C., Lin, C.: LIBSVM: A library for support vector machines (2001), http://www.csie.ntu.edu.tw/~cjlin/libsvm
  12. 12.
    NIST: The NIST Year 2008 Speaker Recognition Evaluation Plan (2008), http://www.nist.gov/speech/tests/sre/2008/sre08_evalplan_release4.pdf
  13. 13.
    Bengio, S., Mariéthoz, J.: A statistical significance test for person authentication. In: Proc. Odyssey, pp. 237–244 (2004)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Mitchell McLaren
    • 1
  • Robbie Vogt
    • 1
  • Brendan Baker
    • 1
  • Sridha Sridharan
    • 1
  1. 1.Speech and Audio Research LaboratoryQUTBrisbaneAustralia

Personalised recommendations