Learning phrase break detection in Thai text-to-speech

One of the crucial problems in developing high quality Thai text-to-speech synthesis is to detect phrase break from Thai texts. Unlike English, Thai has no word boundary delimiter and no punctuation mark at the end of a sentence. It makes the problem more serious. Because when we detect phrase break incorrectly, it is not only producing unnatural speech but also creating the wrong meaning. In this paper, we apply machine learning algorithms namely C4.5 and RIPPER in detecting phrase break. These algorithms can learn useful features for locating a phrase break position. The features which are investigated in our experiments are collocations in different window sizes and the number of syllables before and after a word in question to a phrase break position. We compare the results from C4.5 and RIPPER with a based-line method (Part-of-Speech sequence model). The experiment shows that C4.5 and RIPPER appear to outperform the basedline method and RIPPER performs better accuracy results than C4.5.