Developing a Thai emotional speech corpus

This paper presents a process of developing a Thai emotional speech corpus. The process provides procedures of collecting emotion data from video, voice, and transcription, and that of preparing data for analyzing and annotating. There are two groups (four emotions and twelve emotions) for emotion tagging. The method of emotion analysis and annotation used in this work is also proposed by applying some psychology theory. An annotation tool was developed in order to create this corpus. All procedures of how to use, to analyze, and to tag emotional data are detailed, before the progress of this corpus is given at last.