대화형 음성인터페이스 시스템을 위한 국어 문말 경계억양의 분류

Abstract

The purpose of this research is to provide a classification of sentence-final boundary tones in conversational-style speech for the development of Korean dialog systems. While IP boundary tones in the K-ToBI system are classified based on their shapes, our classification is based on their meaning and function as proposed by Lee(1990, 1996). We propose a more refined classification based on the analysis of speech data containing a wide range of sentence ending types. We then investigate the global distribution of boundary tone types in the speech data as well as their distribution per sentence type.