Comparing to polynomial regression and linear multi-variant regression, support vector regression can not only enhance the generalization ability but also avoid the discontinuousness in spectrum, and the spectrum distance distortion from converted voice t
Comparing to polynomial regression and linear multi-variant regression, support vector regression can not only enhance the generalization ability but also avoid the discontinuousness in spectrum, and the spectrum distance distortion from converted voice t
与多项式回归和线性多变量回归相比，支持向量回归既提高了泛化能力又避免了频谱不连续性，使转换语音与目标语音的频谱距离失真分别减少了33.29%和35.24%。