Rule-based Generation of Phrase Components in Two-step Synthesis of Fundamental Frequency Contours of Mandarin

In this paper, a rule-based method was developed for realizing phrase components in our two-step generation of fundamental frequency (F0) contours of Mandarin. The scheme assumes (logarithmic) F0 contours as superposition of tone components on phrase components, which are further assumed to be responses of phrase commands. In general, possibility of a new phrase command comes higher at deeper syntactic boundaries, but is also affected by the distance from the preceding phrase command. A long interval from preceding phrase command causes a flat F0 contour close to baseline, which is not the case in human speech. In the case of tonal languages such as Mandarin, tone components can be negative. Hence, to give a margin for downward F0 movement, phrase components need to be kept above a certain level, which requires more frequent phrase commands as compared to nontonal languages. Based on these facts, simple rules were constructed for phrase component generation. Speech synthesis was conducted using F0 contours generated by the method. The result of listening test showed a good control of F0 contours being realized by the method.