search

yxlllc / DDSP-SVC

Real-time end-to-end singing voice conversion system based on DDSP (Differentiable Digital Signal Processing)
MIT License
1.87k stars 250 forks source link

DDSP Gui meaning #102

Open benc1221 opened 11 months ago

benc1221 commented 11 months ago

what does kstep and phase vocoder means? is it accent control?

yxlllc commented 11 months ago

kstep is the shallow diffusion step, phase vocoder is for stitching samples in real time, it’s all written in the readme.

benc1221 commented 11 months ago

which leads me to my next question, does the accent bake onto the model while training or is there a setting for that?

yxlllc commented 11 months ago

Accent will affect model performance