Applying Viterbi Bayesian Predictive Classification to Robust Recognition of Continuous Speech