A novel unit selection method for concatenation speech system using similarity measure

This paper presents a new approach to unit selection for corpus-based TTS system, in which the units are selected according to their similarity with synthetic target generated by a parametric synthesizer. In the training stage, a group of classifiers are trained based on human perceptual judgments. The outputs of the classifiers are used to make a distinction rather than using traditional methods such as continuously-valued cost. In order to obtain a better classification result, different combinations of features are tried as input vectors, and the similarity rating is carried out dexterously. Subjective listening tests on a Mandarin female TTS system show that the proposed classifier based speech synthesis system outperforms the traditional unit-selection system.