Digital Speech Processing Homework 1

(Feb 2014 - Mar 2014)

I am asked to implement Hidden Markov Model (HMM) and try different initial state in order to achieve better model performance. From the experiment, one can tell that both the number of iterations and the number of status have a great influence on the final learning result. At the end, my model is able to achieve 0.87 accuracy with 1000 iterations.

Leave a Reply

Your email address will not be published. Required fields are marked *