基于SVR的语音变换映射规则研究
doi: 10.3724/SP.J.1146.2007.00210
The Algorithm of Voice Transformation Based on SVR
-
摘要: 该文介绍了语音变换与支持向量回归(SVR)的基本理论。提出了基于多输出支持向量回归的语音变换特征参数映射规则,并对该映射规则进行了仿真实验。对变换后语音所进行的主客观测试表明,该映射规则对比码书映射和高斯混合模型,能够在参数映射离散性和平滑性之间有效折中,提高语音可懂度。
-
关键词:
- 语音变换; 映射规则; 支持向量回归
Abstract: In this paper, the fundamental theory of voice transformation and Support Vector Regression(SVR) are introduced. The algorithm of voice transformation technology based on multi-output SVR is put forward, and the corresponding simulation experiment is carried out. The subjective test and objective test show that compared with code book mapping and GMM, this mapping rule performs well between discreteness and smoothness of parameter. -
计量
- 文章访问数: 3307
- HTML全文浏览量: 67
- PDF下载量: 1002
- 被引次数: 0
下载:
