Open hpx502766238 opened 1 year ago
w-okada
commented
1 year ago Currently, there are operations that ONNX does not support, so a simple conversion is not possible. It requires fairly advanced mathematical processing. I would greatly appreciate your help.
hpx502766238
commented
1 year ago Currently, there are operations that ONNX does not support, so a simple conversion is not possible. It requires fairly advanced mathematical processing. I would greatly appreciate your help.
Sorry, I'm not capable of algorithmic processing either, but I've found some potentially helpful code from the original author of DiffusionSVC. https://github.com/CNChTu/Diffusion-SVC/blob/Stable/diffusion/onnx_export.py
w-okada
commented
1 year ago I know that. But that's code does not come from author of Diffusion SVC. This author made it for his own software and that exported onnx model can not be used other software, I think. If you want to what is that, you can find his onnx export code in other vc repos such as so-vits-svc, rvc, or others.
https://github.com/CNChTu/Diffusion-SVC/commits/Stable/diffusion/onnx_export.py
Issue Type
Feature Request
vc client version number
MMVCServerSIO_win_onnxdirectML-cuda_v.1.5.3.11.zip
OS
Windows 11
GPU
AMD Radeon RX Vega8
Clear setting
no
Sample model
yes
Input chunk num
yes
Wait for a while
The GUI successfully launched.
read tutorial
yes
Extract files to a new folder.
yes
Voice Changer type
DiffusionSVC,RVC
Model type
pytorch,onnx,f0
Situation
ONNX is much faster, but DiffusionSVC is a PyTorch model. Can you add a feature to export it into ONNX? Additionally, we also need the ONNX version of RMVPE.
application window capture
No response
logs on terminal
vvclient.log