Study of the load balancing in the parallel training for automatic speech recognition

In this paper we propose a parallelization technique of the training phase for the automatic speech recognition using the Hidden Markov Models (HMMs), which improves the load balancing in the previous proposed parallel implementations [1]. This technique is based on an efficient distribution of the vocabulary on processors taking into account, not only the size of the vocabulary, but also the length of each word. In this manner the idle time will be reduced. The experimental results show that good performances can be obtained with this distribution.