A Residue-Based Approach to Text-Independent Speaker Recognition
P. Thévenaz, H. Hügli
Proceedings of the First Swiss Symposium on Pattern Recognition and Computer Vision, Lausanne VD, Swiss Confederation, January 30-31, 1992, pp. 35–41.
The subject of this paper is speaker recognition, which aims at asserting the identity of people on the basis of their voice only. Several techniques are already available; our contribution is to be found in the investigation of features new to this domain, named residue of the linear predictive coding analysis. Our experiments in text-independent mode show that the results obtained by using these features are as good as those obtained by some other classical technique making use of the pitch, while requiring less fine tuning of parameters; hence they can be considered more appealing, because of a more direct implementation.
@INPROCEEDINGS(http://bigwww.epfl.ch/publications/thevenaz9201.html, AUTHOR="Th{\'{e}}venaz, P. and H{\"{u}}gli, H.", TITLE="A Residue-Based Approach to Text-Independent Speaker Recognition", BOOKTITLE="Proceedings of the First Swiss Symposium on Pattern Recognition and Computer Vision", YEAR="1992", editor="", volume="", series="", pages="35--41", address="Lausanne VD, Swiss Confederation", month="January 30-31,", organization="", publisher="", note="")