DDDM-VC: Decoupled Denoising Diffusion Models with Disentangled
 Representation and Prior Mixup for Verifed Robust Voice Conversion  
 
 
 https://ojs.aaai.org/index.php/AAAI/article/view/29740 https://ojs.aaai.org/index.php/AAAI/article/view/29740
https://ojs.aaai.org/index.php/AAAI/article/view/29740 
 
 https://ojs.aaai.org/index.php/AAAI/article/view/29740
https://ojs.aaai.org/index.php/AAAI/article/view/29740 
1.概述
首先,语言有多种属性,如语音文本信息、音调和音色。而在传统的diffusion的生成过程中,所有属性共享参数。因此