Further feature extraction for speaker recognition

This thesis presents a method of extracting a new speaker's voice features for the purpose of synthetically using the voice of the donor speaker. In the small speaker set, it seems good to recognize speaker by their voice by means of the traditional feature extraction. Nevertheless, the performance of recognizer usually depressed owning to the limited feature space, it is hard to deal with the increasing of speaker set to be recognized. Accordingly it proposes a novel feature extraction method, further feature extract (FFE), which is based on some measures such as weight, differential, combination and selection, are taken to explore those voice characteristics that can be used to distinguish different speakers. Experiment based on 138-person YOHO database demonstrates that better performance can be achieved by the proposed method.

[1]  Lawrence R. Rabiner,et al.  A tutorial on hidden Markov models and selected applications in speech recognition , 1989, Proc. IEEE.

[2]  ChiWei Che,et al.  Speaker recognition using HMM with experiments on the yoho database , 1995, EUROSPEECH.

[3]  Biing-Hwang Juang,et al.  Fundamentals of speech recognition , 1993, Prentice Hall signal processing series.

[4]  Joseph P. Campbell,et al.  Testing with the YOHO CD-ROM voice verification corpus , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[5]  B.S. Atal,et al.  Automatic recognition of speakers from their voices , 1976, Proceedings of the IEEE.

[6]  Douglas A. Reynolds,et al.  Robust text-independent speaker identification using Gaussian mixture speaker models , 1995, IEEE Trans. Speech Audio Process..

[7]  Biing-Hwang Juang,et al.  Hidden Markov Models for Speech Recognition , 1991 .

[8]  Jr. J.P. Campbell,et al.  Speaker recognition: a tutorial , 1997, Proc. IEEE.