Abstract
This paper presents a new score normalization method for speaker identification using Gaussian Mixture Model (GMM). The new GMM normalization method has two main advantages: (1) the thresholds are independent to dataset and mapped to the range of \( \left[ {0\% \div 100\% } \right] \) corresponding to your expected accuracy of the system and (2) better performance comparing to common methods. The experimental results suggest the viability of the proposed approach in terms of shortening the development time and providing regular update for model’s parameters.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Roberts, W.J.J., Willmore, J.P.: Automatic speaker recognition using Gaussian mixture models. In: Information, Decision and Control on Proceedings, pp. 465–470. IEEE, Australia (1999)
Li, K.P., Porter, J.: Normalizations and selection of speech segments for speaker recognition scoring. In: International Conference on Acoustics, Speech, and Signal Processing on Proceedings ICASSP, pp. 595–598. IEEE, New York (1988)
Auckenthalera, R., Michael, C., Lloyd-Thomas, H.: Score normalization for text-independent speaker verification systems. Digital Sig. Process. 10, 42–54 (2000)
Reynolds, D.A.: Comparison of background normalization methods for text-independent speaker verification. In: EUROSPEECH on Proceedings, Rhodes, Greece (1997)
Garcia-Romero, D., Gonzalez-Rodriguez, J., Fierrez-Aguilar, J., Ortega-Garcia, J.: U-norm likelihood normalization in pin-based speaker verification systems. In: AVBPA on Proceedings, pp. 208–213. Springer, Heidelberg (2003)
Vandyke, D, Wagner, G.M.: R-Norm: improving inter-speaker variability modelling at the score level via regression score normalization. In: International Speech Communication Association INTERSPEECH, pp. 1–5. ISCA, New York (2013)
Scikit-learn: Machine Learning in Python. http://scikit-learn.org. Accessed 21 June 2018
Luong, C.M.: Vietnam country report 2016: Updated activities on resources development for Vietnamese Speech and NLP. In: Conference of The Oriental Chapter of International Committee for Coordination and Standardization of Speech Databases and Assessment Techniques (O-COCOSDA) on Proceedings. IEEE, Bali, Indonesia (2016)
Dempster, A.P., Laird, N.M., Rubin, D.B.: Maximum likelihood from incomplete data via the em algorithm. J. Royal Stat. Soc. Series B 39(1), 1–38 (1977)
Acknowledgements
This work was supported by Thai Nguyen university of Technology, Vietnam.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this paper
Cite this paper
Nguyen, V.H. (2019). 2S-Norm: A New Score Normalization for a GMM Based Text-Independent Speaker Identification System. In: Fujita, H., Nguyen, D., Vu, N., Banh, T., Puta, H. (eds) Advances in Engineering Research and Application. ICERA 2018. Lecture Notes in Networks and Systems, vol 63. Springer, Cham. https://doi.org/10.1007/978-3-030-04792-4_5
Download citation
DOI: https://doi.org/10.1007/978-3-030-04792-4_5
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-04791-7
Online ISBN: 978-3-030-04792-4
eBook Packages: EngineeringEngineering (R0)