We present a neural prediction system for continuous speaker independent speech recognition. We propose different neural predictors for modeling speech production and discriminative criteria for training. The best system allows to reach 74,9% accuracy on TIMIT which compares well with other state of the art systems. The behavior and performances of this system are then compared with a Hidden Control Neural Network implementation of the predictors.
Keywords: Predictive Neural Networks, Discriminant training, Continuous speech recognition.