A study of prosodic features for Indonesian speech recognition

Abstract

Utterance-type information has been used been used in spoken dialogue system, speech recognition system and translation machine. In a typical spoken dialogue system, a user can ask question or give information to the system. In another side, the spoken dialogue system should be capable of recognizing its user intention to give the correct response to him/her. In this dissertation, the automatic utterance-type recognizer is proposed to distinguish declarative questions from statements in Indonesian speech. Since utterances in these two types have the same words with the same order and differ only in their intonations, their classification requires not only a word recognizer, but also an intonation recognizer. At first, the utterance-type recognizer is designed based on Fujisaki model. The utterance-type recognizer uses a combination of the Fujisaki-model-parameters as the features to recognizt the two utterance type. The best performance of the Fujisaki model based utterance-type recognizer is achieved using a combination of a fraction value of F[subscript b] : F[subscript b]/100 the amplitude of last accent command, and the magnitude of last phrase command as the input of the neural neetworks. However, the Fujisaki parameters extractor is too complicated to be implemented in an automatic recognition system. Therefore, the utterance-type recognizer is developed using the polynomial coefficients of the pitch contours of the sentence's final word. The automatic utterance-type recognizer using polynomial expansion consists of a pitch contour extractor, normalizer, feature extractor, classifier, and an automatic utterance segmentation module. The pitch contour of each utterance type i analyzed to investigate the final word of the two utterance type. To create the automatic utterance segmentation module, an Indonesian acoustic model is designed. The evaluation confirms that the method using the final word and polynomial expansion is effective to distinguish declarative questions and statements in Indonesian speech.