Parameter Trajectory Modeling for Speech Recognition

ABSTRACT
Speech recognition systems live or die by their ability to accurately model acoustics. Our understanding of speech production suggests that there are some underlying constraints on the time course of the spectrum of the speech signal due to the physics of the human articulatory system. State-of-the-art speech recognition systems are extremely overconstrained for their current level of performance. In this talk, we will review some preliminary research on the use of trajectory modeling to improve speech recognition performance.

Institute for Signal and Information Processing
Mississippi State University
Department of Electrical and Computer Engineering
phone/fax: 601-325-3149
email: picone@cavs.msstate.edu
URL: http://cavs.msstate.edu/research/isip/publications/seminars/1997/isip_trajectory_modeling