search

Plachtaa / VITS-fast-fine-tuning

This repo is a pipeline of VITS finetuning for fast speaker adaptation TTS, and many-to-many voice conversion
Apache License 2.0
4.7k stars 705 forks source link

加载vits 模型，使用interfere.exe 出现 size mismatch #259

Open qraccess opened 1 year ago

qraccess commented 1 year ago

Traceback (most recent call last): File "inference.py", line 97, in File "utils.py", line 45, in load_checkpoint File "torch\nn\modules\module.py", line 1672, in load_state_dict self.class.name, "\n\t".join(error_msgs))) RuntimeError: Error(s) in loading state_dict for SynthesizerTrn: size mismatch for enc_p.emb.weight: copying a param with shape torch.Size([52, 192]) from checkpoint, the shape in current model is torch.Size([50, 192]). [9920] Failed to execute script 'inference' due to unhandled exception!

模型来自极客团队训练的派蒙： https://github.com/zixiiu/Digital_Life_Server

yihuitang commented 1 year ago

我也遇到过类似的问题

qraccess commented 1 year ago

有谁知道这个怎么解决？

BigMouseNie commented 1 year ago

解决了吗

BigMouseNie commented 1 year ago

应该是config文件的问题，估计你不是自己训练的

JunhaoHuang0615 commented 1 year ago

这个是因为config文件没有给正确，config文件中的symbol数量与模型不匹配，也许需要找他们更新一下config文件？

WellTung666 commented 1 year ago

自己训练了模型放到数字生命项目中遇到同样的问题，请问如何解决的？谢谢~

WellTung666 commented 1 year ago

这个是因为config文件没有给正确，config文件中的symbol数量与模型不匹配，也许需要找他们更新一下config文件？

请问自己训练的模型需要怎样修改呢？

mikeyang01 commented 10 months ago

+1

xw2018 commented 9 months ago

config文件和预训练模型位置放好之后，训练也有同样的问题。我是训练完成后，把VC_inference.pyL90的--config_dir改成.\configs\modified_finetune_speaker.json，目前没报错