Hearer model based stress prediction for Chinese TTS system

People often feel tired if he/she listens synthesized speech for a long time. This is mainly because synthesized speech is too flat and never stresses the focus. Different to traditional TTS research approach of simulating speaker, this paper does the stress prediction research from the point of the hearer. An ideal hearer model is first proposed to predict the stress distribution based on the hypothesis: people speak with limited stress effort and distribute the limited effort to ensure that the hearer can understand the speaker easily. Then according to the limited research resource, this paper modifies the ideal hearer model and presents a practical model. Experiments show that the stress prediction achieves an acceptable rate of 87.36%.