说话人语音改变技术研究与实现

Research and Implementation of Voice Modification Technique

讨论了一些语音改变技术的应用场合,分析了语音改变的一般方法,并提出了一种新的针对PLAR(pseudologarearatio)曲线进行变换的语音改变方法,从而使说话人的语音特征得到改变。该方法采用了基于音段和听觉的方法来改变语谱特征参数,算法简便,计算量少,便于实时处理。用该算法实现了语音通信中个人隐私的实时保护。

This paper presents a review of some application of voice modification technology and analyses some general methods of it, and proposes a novel approach to change the speaker's identity by transforming the PLAR(pseudo log area ratio)curve. This approach focuses on a segmental, acoustic method to modify the spectral feature parameters .Experiments demonstrate that the algorithm is simple and has light computation load ,therefore it can be implemented easily in a real-time system . This method is put into the application of protecting personal privacy in speech communication.