Plenary 1
Multimodal Speech Perception: A Paradigm for Speech Science

Dominic W. Massaro

Much research has shown that perceivers naturally evaluate and
integrate auditory and visual information in face-to-face speech
perception. Relevant empirical and theoretical research will be
reviewed and evaluated. The research is placed in a general
theoretical framework. Pattern recognition is viewed as central to
cognition, and the perception of speech by eye and ear is deemed as a
prototypical case of pattern recognition. Quantitative models of human
speech perception will be developed and tested against a variety of
empirical results.

We will also cover the technology and science of creating talking
faces. We organize the various approaches that have been taken,
describe our synthesis and animation routines in detail, and provide a
variety of examples. We describe an ongoing series of experiments
aimed at evaluating the realism of visible speech synthesis through
comparison with natural speech

Speechreading is highly functional in a variety of nonoptimal
situations. It follows that the pursuit of visible speech technology
could be of great practical value in many spheres of communication. We
will discuss potential applications of talking faces. Some areas of
application include language training for hearing-impaired
individuals. second language learning, tutoring for poor readers, and
agents in human-machine interactions.

Dr. Dominic W. Massaro       
Department of Psychology               831-459-2330 OFFICE
Social Sciences II                            831-459-2655 LAB     
University of California - Santa Cruz  831-459-5085 MSGS   
Santa Cruz, CA 95064 USA               831-459-3519 FAX