特定说话人之间声学特征参数研究

Study of acoustic parameters between two speakers

特定人语音特征转换技术是实现个性化语音合成的有力工具,无论军用还是民用都具有重要的应用价值。从算法复杂度低、便于应用的角度出发,简要介绍了当今主流特定人语音转换算法的优缺点,着重研究了特定说话人之间基频和共振峰等声学特征参数的内在联系,最后总结了特定人语音特征转换的处理规则。该规则算法实现简单,能够确保转换语音的主要声学特征与目标人相吻合,为研究目标相似度高、实时性好的特定人语音特征转换算法提供了有力的实验依据。

Voice conversion is a new technology of personalized speech synthesis. In this paper, the advantages and disadvantages of mainstream voice conversion algorithm were introduced briefly. The contact between two speakers were studied by experiments significantly. Some rules on voice conversion were summed up. The proposed rules, which can be implemented conveniently, can guarantee high similarity of converted voices and object voices.