An Effective Hybrid Optimization Algorithm for HMM

Hidden Markov model (HMM) is currently the most popular approach to speech recognition. The problem of optimizing model parameters is of great interest to the researchers in this area. The Baum-Welch (BW) algorithm is very popular estimation method due to its reliability and efficiency. However, it is easily trapped in local optimum. particle swarm optimization (PSO) algorithm is a stochastic global optimization technique, but its convergence speed is comparatively slow. With the purpose of overcoming their drawbacks, a new training algorithm based on the PSO algorithm and the BW algorithm (PSOBW) is proposed to train the continuous HMM in continuous speech recognition. This algorithm not only overcomes the shortcoming of the slow convergence speed of the PSO algorithm but also helps the BW algorithm escape from local optimum. The experimental results show that the algorithm is superior to the BW algorithm in the recognition performance.