search

voicepaw / so-vits-svc-fork

so-vits-svc fork with realtime support, improved interface and more features.
Other
8.77k stars 1.18k forks source link

so-vits-svc produce model, can not in use with so-vits-svc-fork #836

Open 729533572 opened 1 year ago

729533572 commented 1 year ago

Describe the bug

RuntimeError: Given groups=1, weight of size [192, 768, 5], expected input[1, 256, 86] to have 768 channels, but got 256 channels instead

To Reproduce

RuntimeError: Given groups=1, weight of size [192, 768, 5], expected input[1, 256, 86] to have 768 channels, but got 256 channels instead

Additional context

RuntimeError: Given groups=1, weight of size [192, 768, 5], expected input[1, 256, 86] to have 768 channels, but got 256 channels instead

Version

latest

Platform

windows11

Code of Conduct

No Duplicate

sbersier commented 1 year ago

The configuration file you are using doesn't correspond to your model.

Try with the so-vits-svc-4.0v1-legacy.json config file. It should be located in: your_path_to_so_vits_svc_fork/venv/lib/python3.10/site-packages/so_vits_svc_fork/preprocessing/config_templates

copy it and paste it in the your_path_to_so_vits_svc_fork/configs folder