Smart Voice系统的设计与实现

Design and implementation of Smart Voice system

针对目前我国语音合成系统合成的语音机器味较浓、语音质量较差的问题,设计和实现了一种中文TTS软件系统Smart Voice,并给出了Smart Voice的调用实现.该系统使用层次化、模块化的程序结构,完成了标记分析、语法分析、延音处理、变调处理等功能.尤其在延音处理器中,创新地采用波形拼接技术创制了波形内部循环的延音模块,使得声音在延音时具有很好的自然度.系统运行表明,Smart Voice发音自然,能为网络、游戏等各种计算机应用提供高速灵活的中文语音支持.

Aiming at the problem that the current speech synthesis systems has strong taste ot speech machine and poor voice quality, Smart Voice, a kind of Chinese T-FS system was designed, and the Smart Voice call implementation was given. The system used a hierarchical, modular program structure, completing the marker analysis, syntax analysis, sustain treatment, tone sandhi processing function. Especially in the sustain processor, waveform concatenation technology innovation was used, creating waveform internal circulation damper module to make a good degree of natural sustain. System operation results showed that Smart Voice sounded natural, and could provide high-speed flexible Chinese speech for the network, games and other computer applications.