Feature Selection Based on Information Theory for Speaker Verification

  • Rafael Fernández
  • Jean-François Bonastre
  • Driss Matrouf
  • José R. Calvo
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5856)

Abstract

Feature extraction/selection is an important stage in every speaker recognition system. Dimension reduction plays a mayor roll due to not only the curse of dimensionality or computation time, but also because of the discriminative relevancy of each feature. The use of automatic methods able to reduce the dimension of the feature space without losing performance is one important problem nowadays. In this sense, a method based on mutual information is studied in order to keep as much discriminative information as possible and the less amount of redundant information. The system performance as a function of the number of retained features is studied.

Keywords

mutual information feature selection speaker verification 

References

  1. 1.
    Campbell, J.P.: Speaker recognition: A tutorial. Proceedings of the IEEE 85(9), 1437–1462 (1997)CrossRefGoogle Scholar
  2. 2.
    Kinnunen, T.: Spectral features for automatic text-independent speaker recognition. Lic. Th., Department of Computer Science, University of Joensuu, Finland (2003)Google Scholar
  3. 3.
    Sambur, M.R.: Selection of acoustic features for speaker identification. IEEE Trans. Acoust. Speech, Signal Processing 23(2), 176–182 (1975)CrossRefGoogle Scholar
  4. 4.
    Aha, D.W., Bankert, R.L.: A comparative evaluation of sequential feature selection algorithms. In: Proceedings of the Fifth International Workshop on Artificial Intelligence and Statistics, pp. 1–7. Springer, Heidelberg (1995)Google Scholar
  5. 5.
    Fauve, B.: Tackling Variabilities in Speaker Verification with a Focus on Short Durations. PhD thesis, School of Engineering Swansea University (2009)Google Scholar
  6. 6.
    Peng, H., Long, F., Ding, C.: Feature selection based on mutual information: Criteria of max-dependency, max-relevance and min-redundancy. IEEE Trans. Patt. Anal. and Mach. Intel. 27(8), 1226–1238 (2005)CrossRefGoogle Scholar
  7. 7.
    Fernández, R., Montalvo, A., Calvo, J.R., Hernández, G.: Selection of the best wavelet packet nodes based on mutual information for speaker identification. In: Ruiz-Shulcloper, J., Kropatsch, W.G. (eds.) CIARP 2008. LNCS, vol. 5197, pp. 78–85. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  8. 8.
    Bonastre, J.F., et al.: ALIZE/spkdet: a state-of-the-art open source software for speaker recognition, Odyssey, Stellenbosch, South Africa (January 2008)Google Scholar
  9. 9.
    Torkkola, K., Campbell, W.M.: Mutual information in learning feature transformations. In: Proc. Int. Conf. on Mach. Learning, San Francisco, CA, USA, pp. 1015–1022. Morgan Kaufmann Publishers Inc., San Francisco (2000)Google Scholar
  10. 10.
    Cover, T.M., Thomas, J.A.: Elements of Information Theory. Wiley-Interscience, Hoboken (1991)MATHCrossRefGoogle Scholar
  11. 11.
    Lu, X., Dang, J.: Dimension reduction for speaker identification based on mutual information. In: Interspeech, pp. 2021–2024 (2007)Google Scholar
  12. 12.
  13. 13.
    Gravier, G.: SPRO: a free speech signal processing toolkit, http://www.irisa.fr/metiss/guig/spro/
  14. 14.
    Martin, A., Doddington, G., Kamm, T., Ordowski, M., Przybocki, M.: The DET curve in assessment of detection task performance. In: Eurospeech, Rhodes, Greece, September 1997, pp. 1895–1898 (1997)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Rafael Fernández
    • 1
    • 2
  • Jean-François Bonastre
    • 2
  • Driss Matrouf
    • 2
  • José R. Calvo
    • 1
  1. 1.Advanced Technologies Application CenterHavanaCuba
  2. 2.Laboratoire d’Informatique d’AvignonUAPVFrance

Personalised recommendations