“…We use the semantic primitive specifications with corresponding acoustic features of the first step to resynthesize the expressive speech utterances, and ask human subjects to evaluate the effectiveness of the resynthesized utterances. This approach includes three techniques: (1) a voice-morphed technique for rule verification, (2) expression of the relationships built by step 1 as rules, and (3) examination of various types of human perception of emotion as well as the intensity of the perceptions (Huang and Akagi, 2005).…”