【Actual Combat】Experience SadTalker

Paper http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_SadTalker_Learning_Realistic_3D_Motion_Coefficients_for_Stylized_Audio-Driven_Single_CVPR_2023_paper.pdf

 github

GitHub - OpenTalker/SadTalker: [CVPR 2023] SadTalker：Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation

It's been a while since this stuff came out. I have been waiting for the training code, but it seems that there is no end in sight. Let me make a simple environmental record first, and then record it together with the paper when the training code comes out later.

1. Experience

git clone https://github.com/Winfredy/SadTalker.git

cd SadTalker 

conda create -n sadtalker python=3.8

conda activate sadtalker

pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu113

conda install ffmpeg

pip install -r requirements.txt

### tts is optional for gradio demo. 
### pip install TTS

It went very smoothly, basically there is no problem, but it takes a long time to download the checkpoint

I lost all these two weights on Baidu Netdisk 

Link: https://pan.baidu.com/s/18ssIL7vrRg_mcov5H6hEQA?pwd=9xkm Extraction code: 9xkm 
--Share from Baidu Netdisk super member v2

2. Training

Waiting for the training code....