This paper discusses a speech recognition system using the continuous mixture density Hidden Markov Model. We use a nonlinear programming method to proceed time warping about feature vector. Moreover
we present a method based on speech knowledge to initialize model.