Text this: On the use of visual speech information and generalized fuzzy hidden Markov models for speech recognition /