Conversion fails instantly despite successful training and indexing

Hi all, I could really use some help.

I’ve successfully trained my model and generated the corresponding .index file using train-index-v2.py. However, no matter what I try, the audio conversion fails instantly with the following error:

2025-06-30 01:56:59 | WARNING | infer.modules.vc.modules | Traceback (most recent call last):
  File "C:\\Users\\mys\\Desktop\\Retrieval-based-Voice-Conversion-WebUI\\infer\\modules\\vc\\modules.py", line 172, in vc_single
    self.hubert_model = load_hubert(self.config)
  File "C:\\Users\\mys\\Desktop\\Retrieval-based-Voice-Conversion-WebUI\\infer\\modules\\vc\\utils.py", line 23, in load_hubert
    models, _, _ = checkpoint_utils.load_model_ensemble_and_task(
  File "C:\\Users\\mys\\.venv_py310\\lib\\site-packages\\fairseq\\checkpoint_utils.py", line 425, in load_model_ensemble_and_task
    state = load_checkpoint_to_cpu(filename, arg_overrides)
  File "C:\\Users\\mys\\.venv_py310\\lib\\site-packages\\fairseq\\checkpoint_utils.py", line 317, in load_checkpoint_to_cpu
    if "args" in state and state["args"] is not None and arg_overrides is not None:
TypeError: argument of type 'NoneType' is not iterable

Then Gradio throws this:

AttributeError: 'NoneType' object has no attribute 'tobytes'
It seems like HuBERT fails to load but I’ve double-checked that the hubert.pt file is definitely present in assets/hubert.

Has anyone run into this before? Any ideas how I can fix this?

Thanks in advance!

Start in infer/modules/vc/modules.py at vc_single and infer/modules/vc/utils.py at load_hubert, then inspect the fairseq checkpoint-loading traceback using the hubert.pt file under assets/hubert. Reproduce the conversion failure and trace why model loading returns None; done means conversion produces audio without the Gradio NoneType error.