A Speech Synthesis System with Emotion for Assisting Communication

The authors have proposed a method of generating synthetic speech with emotion by creating three copora of emotional speech for use with CHATR [1][2], the concatenative speech synthesis system developed at ATR [3][4]. The corpora express joy, anger and sadness. For the previous trial, speech corpora were made with female voice. Having added speech corpora of male voice, the authors developed a prototype of a speech synthesis system with emotion, “CHATAKO,” which works on a notebook PC in Japanese language environment. The ultimate goal of this system is to assist people with communication problems. In this paper, first, design principles are stated, followed by system configuration and the evaluation on the synthetic speech with emotion generated by our method.