基于TD-PSOLA算法的语音合成方法研究

Research on the Method of Speech Synthesis Based on TD-PSOLA

摘要:

介绍了一种采用TD-PSOLA合成语音的新方法,针对采用自相关法提取基音周期的缺点,该方法运用中心削波法降低计算短时信号的自相关函数的计算量。经过仿真验证,该方法大大降低了原算法的复杂度且在一定的范围内能合成高质量的语音,具有很强的工程价值。

Abstract:

A new method of speech synthesis adopting TD-PSOLA was introduced. According to the disadvantage of the method that using auto-correlation function to abstract the periodic time of pitch,this method used center-clipping method to reduce the computational complexity of calculating auto-correlation functions of the fast signal. The method can largely reduce complexity of the algorithmic, and it can synthesize high-grade sounds within certain limits. It has great construction value.