Application of ANN and HMM to Automatic Speaker Verification

This paper describes the development and implementation of a speaker verification system, applied as a telephony security tool. This software was made using two of the most popular techniques in the commercial recognition systems, techniques based in Hidden Markov Models (HMM) and Artificial Neural Networks (ANN) theory. First of all, the qualities of the system were chosen. Then, the different parts that make up the verification system (speech characteristics extraction, pattern recognition, decision rules, among others) were developed. Therefore, the system was tested in order to adjust design parameters and to evaluate its performance. Finally the verification system took part in a interactive answering machine, using a Intel Dialogic ProLine/2V telephony card for this purpose.

[1]  Sadaoki Furui,et al.  An Overview of Speaker Recognition Technology , 1996 .

[2]  P. J. Werbos,et al.  Backpropagation: past and future , 1988, IEEE 1988 International Conference on Neural Networks.

[3]  S. Liu,et al.  A practical guide to biometric security technology , 2001 .

[4]  A.E. Rosenberg,et al.  Automatic speaker verification: A review , 1976, Proceedings of the IEEE.

[5]  Richard P. Lippmann,et al.  An introduction to computing with neural nets , 1987 .