A Targets-based Superpositional Model of Fundamental Frequency Contours Applied to HMM-based Speech Synthesis

Superpositional model of fundamental frequency (F-0) contours as suggested by the Fujisaki model can well represent F-0 movements of speech keeping a clear relation with linguistic information of utterances. Therefore, improvement of HMM-based speech synthesis is expected by using the merit of superpositional model. In this paper, a targets-based superpositional model is proposed in the light of the Fujisaki model. Here, both accent and phrase components are parameterized by respectively defined low and high targets which allow flexible interaction between accent and phrase components. Due to the flexible interaction, the new method consistently treats such complex Fo movements as low digging, varying declination, and final lowering by simply adjusting parameter values. This facilitates extraction of the model parameters from observed F-0 contours, which is one of major problems preventing the use of the Fujisaki model. Extraction of the target parameters is evaluated for a Japanese speech corpus and the F-0 contours generated by the model are used for HMM training instead of the original. Listening test of synthetic speech indicates significant improvements in speech quality. Micro-prosodic effects are also investigated. Results show that adding the micro-prosody to the generated F-0 contours does not significantly improve speech quality.