What can visual speech synthesis tell visual speech recognition?

We consider the problem of speech recognition given visual and auditory information, and discuss some of the ways that speech synthesis can provide assistance. Three possible contributions of synthetic visual speech are discussed. First, the use of synthetic speech to study human speech perception, second, the use of speech synthesis techniques to instantiate models of human speech production, and third, the use of these production models to help guide automatic speech recognition. Finally, we consider the reverse relationship. How can the techniques of automatic speech recognition assist in better visual speech synthesis?