Neural networks with random letter codes for text-to-phoneme mapping and small training dictionary
Enikö Beatrice Bilcu, Jaakko T. Astola · 2006
In this paper we address the problem of text-to-phoneme (TTP) mapping implemented by neural networks. One im-portant disadvantage of the neural networks is the conver-gence interval which can be in some situations very large. Even when the neural networks are trained in off line mode a shorter convergence interval would be of interest due to var-ious reasons. In the TTP mapping, decreasing the number of necessary iterations is equivalent to relaxing the require-ments for the dictionary size. In this paper, we show that proper letter encoding can increase the convergence speed of the multilayer perceptron neural network for the task of TTP mapping. Experimental results that compare the perfor-mance of several techniques that speed-up the convergence of the multilayer perceptron, in the context of TTP mapping are also presented. 1.