A hybrid RBF-HMM system for continuous speech recognition

A hybrid system for continuous speech recognition, consisting of a neural network with radial basis functions and hidden Markov models is described in this paper together with discriminant training techniques. Initially the neural net is trained to approximate a-posteriori probabilities of single HMM states. These probabilities are used by the Viterbi algorithm to calculate the total scores for the individual hybrid phoneme models. The final training of the hybrid system is based on the `minimum classification error´ objective function, which approximates the misclassification rate of the hybrid classifier, and the `generalized probabilistic descent´ algorithm. The hybrid system was used in continuous speech recognition experiments with phoneme units and shows about 63.8% phoneme recognition rate in a speaker-independent task