Wav2Lip

Wav2Lip copied to clipboard

anaconda3\envs\wav2lip\lib\site-packages\torch\serialization.py", line 1264, in _legacy_load typed_storage._untyped_storage._set_from_file( RuntimeError: unexpected EOF, expected 740598 more bytes. The file might be corrupted.

aunirban1234

Run inference.py on windows 11, stucks after readings files

5

- Used python 3.6 \ 3.8 \ 3.10 - torch 1.12 \ for cuda 117 and 118 and others - On linux python3.10 with torch==2.0.1 works fine, without stucks using...

DAKExDUCK

As W2L is quite old and its dependencies can not easily be ported to Python3.11 I wrote a small wrapper that starts a flask server serving W2L. Additionally, as the...

sammyf

What causes black spots on the face of images generated by models trained on my customer dataset？

7

What causes black spots on the face of images generated by models trained on my customer dataset？ 

WYanNuo

If the number `i` of the image is incremented before the test, there may be a jump in the image numbering.

laurentbenaroya

TypeError: guvectorize() missing 1 required positional argument: 'signature'

2

我用的是python3.6.5 \python3.6.5\lib\site-packages\resampy\interpn.py", line 75, in nopython=True, TypeError: guvectorize() missing 1 required positional argument: 'signature'

jack-willson

I have posted the issue in the wav2lip_288 repo, please check detail in: https://github.com/primepake/wav2lip_288x288/issues/98#issue-2055559708

ChengsongLu

wav2lip_train training error：Expected more than 1 value per channel when training, got input size torch.Size([1, 512, 1, 1])

6

 Using my personal data set, the batch size in hparams.py is 16

cncbec

python preprocess.py --ngpu 1 --batch_size 2 --data_root .\filelists\ --preprocessed_root .\raw_dataprocessed Started processing for .\filelists\ with 1 GPUs 0%| | 0/4 [00:00

dididiskq

The amplitude of the mouth shape becomes smaller with custom dataset

7

Guys, I found the amplitude of the mouth shape becomes smaller while use custom dataset，but the pronunciation is correct. My dataset has hundreds of characters ID，and the sync expert loss...

gs-ren