Combinaison de codeurs par algorithme génétique : Application à la vérification du locuteur

Speaker verification aim at recognize the identity of a speaker based on his voice. Applications include security access control, banking phone transaction, surveillance, audio-indexing and forensic speaker recognition. Our study deals with speech feature extraction witch aim at extract speaker discriminative information from the voice signal. In this paper, we propose to use a genetic algorithm to design a feature extraction system adapted to the speaker recognition task. Result show that the proposed method improves significantly the system's performances on the 2005 Nist SRE database.

[1]  Mohamed Chetouani,et al.  Filter Bank Design for Speaker Diarization Based on Genetic Algorithms , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.

[2]  Shigeru Katagiri,et al.  Handbook of Neural Networks for Speech Processing , 2000 .

[3]  Samy Bengio,et al.  Spectral Subband Centroids as Complementary Features for Speaker Authentication , 2004, ICBA.

[4]  Luis Javier Rodríguez-Fuentes,et al.  Feature Selection Based on Genetic Algorithms for Speaker Recognition , 2006, 2006 IEEE Odyssey - The Speaker and Language Recognition Workshop.

[5]  Kari Torkkola,et al.  Feature Extraction by Non-Parametric Mutual Information Maximization , 2003, J. Mach. Learn. Res..

[6]  Zhaohui Wu,et al.  Further feature extraction for speaker recognition , 2003, SMC'03 Conference Proceedings. 2003 IEEE International Conference on Systems, Man and Cybernetics. Conference Theme - System Security and Assurance (Cat. No.03CH37483).

[7]  Chin-Teng Lin,et al.  GA-based noisy speech recognition using two-dimensional cepstrum , 2000, IEEE Trans. Speech Audio Process..

[8]  Marcos Faúndez-Zanuy,et al.  Non-linear Speech Feature Extraction for Phoneme Classification and Speaker Recognition , 2004, Summer School on Neural Networks.