PaperPower exponential densities for the training and classification of acoustic feature vectors in speech recognition
