Sunteți pe pagina 1din 1

Matlab code for recognizing spoken words.

Create one folder for each word and pl ace the training files inside. One hidden Markov model will be trained for each word. The observable output from each hidden state is modeled as a multivariate Gaussi an distribution with mean and covariance depending on the state. The forward alg orithm is used for classification, and the Baum-Welch expectation-maximization a lgorithm is used for training. The approach is explained by L. R. Rabiner, "A tu torial on hidden Markov models and selected applications in speech recognition," Proceedings of the IEEE, vol. 77, pp. 257 286, Feb 1989. Created by Hkon Sandsmark.

S-ar putea să vă placă și