摘要 |
PURPOSE: To generate a synthesized voice having a more natural pitch pattern by inputting a phonemic symbol string and accent information and determining the position of the accent nucleus and sizes of accent components in accordance with the input speaking speed. CONSTITUTION: A phonemic symbol string, delimiting symbols indicating word boundaries, etc., which express a sentence for voice synthesis are inputted from a character string input terminal 11, and the speaking speed is inputted from a speaking speed input terminal 12 and s sent to a duration determination part 14 and an accent nucleus determination part 15. The accent nucleus determination part 15 determines generation/non-generation of auxiliary accents, coupling of accent components, movement of the accent nucleus accompanied with this coupling, and sizes of accent components in accordance with the designated speaking speed and sends them to a pitch pattern generation part 16. This part 16 generates a pitch pattern of a synthesized voice based on determined accent nucleus position and accent component sizes. Thus, the synthesized voice having a natural pitch pattern is generated even if the speaking speed is changed. |