Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

一个奇怪的问题 #97

Open
JSYdiyishuai opened this issue Dec 6, 2024 · 4 comments
Open

一个奇怪的问题 #97

JSYdiyishuai opened this issue Dec 6, 2024 · 4 comments

Comments

@JSYdiyishuai
Copy link

使用不同音频测试,口型是训练音频的口型,请问一下这是什么问题?使用过hubert、wenet、和synctalk的ave提取音频特征,发现都有这个问题,训练loss都很正常,也试过关闭sync_loss。

@Liwenru88
Copy link

同样的问题 训练集和验证集都很正常 但是用其它音频 口型就对不上

@JSYdiyishuai
Copy link
Author

同样的问题 训练集和验证集都很正常 但是用其它音频 口型就对不上

另外我做了个测试,pred = net(img_concat_T, audio_feat)[0],audio_feat无论是什么全0或全1,似乎都影响不了最后生成的口型

@anliyuan
Copy link
Owner

anliyuan commented Dec 6, 2024

目前我还没遇到过这个问题,可以详细写一下复现流程

@JSYdiyishuai
Copy link
Author

目前我还没遇到过这个问题,可以详细写一下复现流程

流程基本如下:
1.只是clone了这个仓库,没有做任何修改
2.cd data_utils;python process.py ./my_video --asr hubert
3. cd ..; python train.py --dataset_dir ./my_video/ --save_dir ./checkpoints/1/ --asr hubert
4. python inference.py --asr hubert --dataset ./my_video/ --audio_feat new.npy --save_path ./tmp.mp4 --checkpoint ./checkpoints/1/195.pth
5. ffmpeg -i tmp.mp4 -i new.wav -c:v libx264 -c:a aac tmp_aud.mp4

可以提供你的训练视频和权重吗?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants