DiffTalk

The pytorch implementation for our CVPR2023 paper "DiffTalk: Crafting Diffusion Models for Generalized Audio-Driven Portraits Animation".

[Project] [Paper] [Video Demo]

Requirements

python 3.7.0
pytorch 1.10.0
pytorch-lightning 1.2.5
torchvision 0.11.0
pytorch-lightning==1.2.5

For more details, please refer to the requirements.txt. We conduct the experiments with 8 NVIDIA 3090Ti GPUs.

Put the first stage model to ./models.

Dataset

Please download the HDTF dataset for training and test, and process the dataset as following.

Data Preprocessing:

Set all videos to 25 fps.
Extract the audio signals and facial landmarks.
Put the processed data in ./data/HDTF, and construct the data directory as following.
Constract the data_train.txt and data_test.txt as following.

./data/HDTF:

|——data/HDTF
   |——images
      |——0_0.jpg
      |——0_1.jpg
      |——...
      |——N_M.bin
   |——landmarks
      |——0_0.lmd
      |——0_1.lmd
      |——...
      |——N_M.lms
   |——audio_smooth
      |——0_0.npy
      |——0_1.npy
      |——...
      |——N_M.npy

./data/data_train(test).txt:

0_0
0_1
0_2
...
N_M

N is the total number of classes, and M is the class size.

Training

sh run.sh

Test

sh inference.sh

Weakness

The DiffTalk models talking head generation as an iterative denoising process, which needs more time to synthesize a frame compared with most GAN-based approaches. This is also a common problem of LDM-based works.
The model is trained on the HDTF dataset, and it sometimes fails on some identities from other datasets.
When driving a portrait with more challenging cross-identity audio, the audio-lip synchronization of the synthesized video is slightly inferior to the ones under self-driven setting.
During inference, the network is also sensitive to the mask shape in z_T , where the mask needs to cover the mouth region completely and its shape cannot leak any lip shape information.

Acknowledgement

This code is built upon the publicly available code latent-diffusion. Thanks the authors of latent-diffusion for making their excellent work and codes publicly available.

Citation

Please cite the following paper if you use this repository in your research.

@inproceedings{shen2023difftalk,
   author={Shen, Shuai and Zhao, Wenliang and Meng, Zibin and Li, Wanhua and Zhu, Zheng and Zhou, Jie and Lu, Jiwen},
   title={DiffTalk: Crafting Diffusion Models for Generalized Audio-Driven Portraits Animation},
   booktitle={CVPR},
   year={2023}
}

syunar/DiffTalk