开发组账号 commited on
Commit
c5bbdcc
β€’
1 Parent(s): 7318c8e
Files changed (2) hide show
  1. README_T.md +152 -0
  2. requirements.txt +3 -1
README_T.md ADDED
@@ -0,0 +1,152 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # **Wav2Lip**: *Accurately Lip-syncing Videos In The Wild*
2
+
3
+ For commercial requests, please contact us at radrabha.m@research.iiit.ac.in or prajwal.k@research.iiit.ac.in. We have an HD model ready that can be used commercially.
4
+
5
+ This code is part of the paper: _A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild_ published at ACM Multimedia 2020.
6
+
7
+ [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/a-lip-sync-expert-is-all-you-need-for-speech/lip-sync-on-lrs2)](https://paperswithcode.com/sota/lip-sync-on-lrs2?p=a-lip-sync-expert-is-all-you-need-for-speech)
8
+ [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/a-lip-sync-expert-is-all-you-need-for-speech/lip-sync-on-lrs3)](https://paperswithcode.com/sota/lip-sync-on-lrs3?p=a-lip-sync-expert-is-all-you-need-for-speech)
9
+ [![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/a-lip-sync-expert-is-all-you-need-for-speech/lip-sync-on-lrw)](https://paperswithcode.com/sota/lip-sync-on-lrw?p=a-lip-sync-expert-is-all-you-need-for-speech)
10
+
11
+ |πŸ“‘ Original Paper|πŸ“° Project Page|πŸŒ€ Demo|⚑ Live Testing|πŸ“” Colab Notebook
12
+ |:-:|:-:|:-:|:-:|:-:|
13
+ [Paper](http://arxiv.org/abs/2008.10010) | [Project Page](http://cvit.iiit.ac.in/research/projects/cvit-projects/a-lip-sync-expert-is-all-you-need-for-speech-to-lip-generation-in-the-wild/) | [Demo Video](https://youtu.be/0fXaDCZNOJc) | [Interactive Demo](https://bhaasha.iiit.ac.in/lipsync) | [Colab Notebook](https://colab.research.google.com/drive/1tZpDWXz49W6wDcTprANRGLo2D_EbD5J8?usp=sharing) /[Updated Collab Notebook](https://colab.research.google.com/drive/1IjFW1cLevs6Ouyu4Yht4mnR4yeuMqO7Y#scrollTo=MH1m608OymLH)
14
+
15
+ <img src="https://drive.google.com/uc?export=view&id=1Wn0hPmpo4GRbCIJR8Tf20Akzdi1qjjG9"/>
16
+
17
+ ----------
18
+ **Highlights**
19
+ ----------
20
+ - Weights of the visual quality disc has been updated in readme!
21
+ - Lip-sync videos to any target speech with high accuracy :100:. Try our [interactive demo](https://bhaasha.iiit.ac.in/lipsync).
22
+ - :sparkles: Works for any identity, voice, and language. Also works for CGI faces and synthetic voices.
23
+ - Complete training code, inference code, and pretrained models are available :boom:
24
+ - Or, quick-start with the Google Colab Notebook: [Link](https://colab.research.google.com/drive/1tZpDWXz49W6wDcTprANRGLo2D_EbD5J8?usp=sharing). Checkpoints and samples are available in a Google Drive [folder](https://drive.google.com/drive/folders/1I-0dNLfFOSFwrfqjNa-SXuwaURHE5K4k?usp=sharing) as well. There is also a [tutorial video](https://www.youtube.com/watch?v=Ic0TBhfuOrA) on this, courtesy of [What Make Art](https://www.youtube.com/channel/UCmGXH-jy0o2CuhqtpxbaQgA). Also, thanks to [Eyal Gruss](https://eyalgruss.com), there is a more accessible [Google Colab notebook](https://j.mp/wav2lip) with more useful features. A tutorial collab notebook is present at this [link](https://colab.research.google.com/drive/1IjFW1cLevs6Ouyu4Yht4mnR4yeuMqO7Y#scrollTo=MH1m608OymLH).
25
+ - :fire: :fire: Several new, reliable evaluation benchmarks and metrics [[`evaluation/` folder of this repo]](https://github.com/Rudrabha/Wav2Lip/tree/master/evaluation) released. Instructions to calculate the metrics reported in the paper are also present.
26
+
27
+ --------
28
+ **Disclaimer**
29
+ --------
30
+ All results from this open-source code or our [demo website](https://bhaasha.iiit.ac.in/lipsync) should only be used for research/academic/personal purposes only. As the models are trained on the <a href="http://www.robots.ox.ac.uk/~vgg/data/lip_reading/lrs2.html">LRS2 dataset</a>, any form of commercial use is strictly prohibhited. For commercial requests please contact us directly!
31
+
32
+ Prerequisites
33
+ -------------
34
+ - `Python 3.6`
35
+ - ffmpeg: `sudo apt-get install ffmpeg`
36
+ - Install necessary packages using `pip install -r requirements.txt`. Alternatively, instructions for using a docker image is provided [here](https://gist.github.com/xenogenesi/e62d3d13dadbc164124c830e9c453668). Have a look at [this comment](https://github.com/Rudrabha/Wav2Lip/issues/131#issuecomment-725478562) and comment on [the gist](https://gist.github.com/xenogenesi/e62d3d13dadbc164124c830e9c453668) if you encounter any issues.
37
+ - Face detection [pre-trained model](https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth) should be downloaded to `face_detection/detection/sfd/s3fd.pth`. Alternative [link](https://iiitaphyd-my.sharepoint.com/:u:/g/personal/prajwal_k_research_iiit_ac_in/EZsy6qWuivtDnANIG73iHjIBjMSoojcIV0NULXV-yiuiIg?e=qTasa8) if the above does not work.
38
+
39
+ Getting the weights
40
+ ----------
41
+ | Model | Description | Link to the model |
42
+ | :-------------: | :---------------: | :---------------: |
43
+ | Wav2Lip | Highly accurate lip-sync | [Link](https://iiitaphyd-my.sharepoint.com/:u:/g/personal/radrabha_m_research_iiit_ac_in/Eb3LEzbfuKlJiR600lQWRxgBIY27JZg80f7V9jtMfbNDaQ?e=TBFBVW) |
44
+ | Wav2Lip + GAN | Slightly inferior lip-sync, but better visual quality | [Link](https://iiitaphyd-my.sharepoint.com/:u:/g/personal/radrabha_m_research_iiit_ac_in/EdjI7bZlgApMqsVoEUUXpLsBxqXbn5z8VTmoxp55YNDcIA?e=n9ljGW) |
45
+ | Expert Discriminator | Weights of the expert discriminator | [Link](https://iiitaphyd-my.sharepoint.com/:u:/g/personal/radrabha_m_research_iiit_ac_in/EQRvmiZg-HRAjvI6zqN9eTEBP74KefynCwPWVmF57l-AYA?e=ZRPHKP) |
46
+ | Visual Quality Discriminator | Weights of the visual disc trained in a GAN setup | [Link](https://iiitaphyd-my.sharepoint.com/:u:/g/personal/radrabha_m_research_iiit_ac_in/EQVqH88dTm1HjlK11eNba5gBbn15WMS0B0EZbDBttqrqkg?e=ic0ljo) |
47
+
48
+ Lip-syncing videos using the pre-trained models (Inference)
49
+ -------
50
+ You can lip-sync any video to any audio:
51
+ ```bash
52
+ python inference.py --checkpoint_path <ckpt> --face <video.mp4> --audio <an-audio-source>
53
+ ```
54
+ The result is saved (by default) in `results/result_voice.mp4`. You can specify it as an argument, similar to several other available options. The audio source can be any file supported by `FFMPEG` containing audio data: `*.wav`, `*.mp3` or even a video file, from which the code will automatically extract the audio.
55
+
56
+ ##### Tips for better results:
57
+ - Experiment with the `--pads` argument to adjust the detected face bounding box. Often leads to improved results. You might need to increase the bottom padding to include the chin region. E.g. `--pads 0 20 0 0`.
58
+ - If you see the mouth position dislocated or some weird artifacts such as two mouths, then it can be because of over-smoothing the face detections. Use the `--nosmooth` argument and give another try.
59
+ - Experiment with the `--resize_factor` argument, to get a lower resolution video. Why? The models are trained on faces which were at a lower resolution. You might get better, visually pleasing results for 720p videos than for 1080p videos (in many cases, the latter works well too).
60
+ - The Wav2Lip model without GAN usually needs more experimenting with the above two to get the most ideal results, and sometimes, can give you a better result as well.
61
+
62
+ Preparing LRS2 for training
63
+ ----------
64
+ Our models are trained on LRS2. See [here](#training-on-datasets-other-than-lrs2) for a few suggestions regarding training on other datasets.
65
+ ##### LRS2 dataset folder structure
66
+
67
+ ```
68
+ data_root (mvlrs_v1)
69
+ β”œβ”€β”€ main, pretrain (we use only main folder in this work)
70
+ | β”œβ”€β”€ list of folders
71
+ | β”‚ β”œβ”€β”€ five-digit numbered video IDs ending with (.mp4)
72
+ ```
73
+
74
+ Place the LRS2 filelists (train, val, test) `.txt` files in the `filelists/` folder.
75
+
76
+ ##### Preprocess the dataset for fast training
77
+
78
+ ```bash
79
+ python preprocess.py --data_root data_root/main --preprocessed_root lrs2_preprocessed/
80
+ ```
81
+ Additional options like `batch_size` and number of GPUs to use in parallel to use can also be set.
82
+
83
+ ##### Preprocessed LRS2 folder structure
84
+ ```
85
+ preprocessed_root (lrs2_preprocessed)
86
+ β”œβ”€β”€ list of folders
87
+ | β”œβ”€β”€ Folders with five-digit numbered video IDs
88
+ | β”‚ β”œβ”€β”€ *.jpg
89
+ | β”‚ β”œβ”€β”€ audio.wav
90
+ ```
91
+
92
+ Train!
93
+ ----------
94
+ There are two major steps: (i) Train the expert lip-sync discriminator, (ii) Train the Wav2Lip model(s).
95
+
96
+ ##### Training the expert discriminator
97
+ You can download [the pre-trained weights](#getting-the-weights) if you want to skip this step. To train it:
98
+ ```bash
99
+ python color_syncnet_train.py --data_root lrs2_preprocessed/ --checkpoint_dir <folder_to_save_checkpoints>
100
+ ```
101
+ ##### Training the Wav2Lip models
102
+ You can either train the model without the additional visual quality disriminator (< 1 day of training) or use the discriminator (~2 days). For the former, run:
103
+ ```bash
104
+ python wav2lip_train.py --data_root lrs2_preprocessed/ --checkpoint_dir <folder_to_save_checkpoints> --syncnet_checkpoint_path <path_to_expert_disc_checkpoint>
105
+ ```
106
+
107
+ To train with the visual quality discriminator, you should run `hq_wav2lip_train.py` instead. The arguments for both the files are similar. In both the cases, you can resume training as well. Look at `python wav2lip_train.py --help` for more details. You can also set additional less commonly-used hyper-parameters at the bottom of the `hparams.py` file.
108
+
109
+ Training on datasets other than LRS2
110
+ ------------------------------------
111
+ Training on other datasets might require modifications to the code. Please read the following before you raise an issue:
112
+
113
+ - You might not get good results by training/fine-tuning on a few minutes of a single speaker. This is a separate research problem, to which we do not have a solution yet. Thus, we would most likely not be able to resolve your issue.
114
+ - You must train the expert discriminator for your own dataset before training Wav2Lip.
115
+ - If it is your own dataset downloaded from the web, in most cases, needs to be sync-corrected.
116
+ - Be mindful of the FPS of the videos of your dataset. Changes to FPS would need significant code changes.
117
+ - The expert discriminator's eval loss should go down to ~0.25 and the Wav2Lip eval sync loss should go down to ~0.2 to get good results.
118
+
119
+ When raising an issue on this topic, please let us know that you are aware of all these points.
120
+
121
+ We have an HD model trained on a dataset allowing commercial usage. The size of the generated face will be 192 x 288 in our new model.
122
+
123
+ Evaluation
124
+ ----------
125
+ Please check the `evaluation/` folder for the instructions.
126
+
127
+ License and Citation
128
+ ----------
129
+ Theis repository can only be used for personal/research/non-commercial purposes. However, for commercial requests, please contact us directly at radrabha.m@research.iiit.ac.in or prajwal.k@research.iiit.ac.in. We have an HD model trained on a dataset allowing commercial usage. The size of the generated face will be 192 x 288 in our new model. Please cite the following paper if you use this repository:
130
+ ```
131
+ @inproceedings{10.1145/3394171.3413532,
132
+ author = {Prajwal, K R and Mukhopadhyay, Rudrabha and Namboodiri, Vinay P. and Jawahar, C.V.},
133
+ title = {A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild},
134
+ year = {2020},
135
+ isbn = {9781450379885},
136
+ publisher = {Association for Computing Machinery},
137
+ address = {New York, NY, USA},
138
+ url = {https://doi.org/10.1145/3394171.3413532},
139
+ doi = {10.1145/3394171.3413532},
140
+ booktitle = {Proceedings of the 28th ACM International Conference on Multimedia},
141
+ pages = {484–492},
142
+ numpages = {9},
143
+ keywords = {lip sync, talking face generation, video generation},
144
+ location = {Seattle, WA, USA},
145
+ series = {MM '20}
146
+ }
147
+ ```
148
+
149
+
150
+ Acknowledgements
151
+ ----------
152
+ Parts of the code structure is inspired by this [TTS repository](https://github.com/r9y9/deepvoice3_pytorch). We thank the author for this wonderful code. The code for Face Detection has been taken from the [face_alignment](https://github.com/1adrianb/face-alignment) repository. We thank the authors for releasing their code and models. We thank [zabique](https://github.com/zabique) for the tutorial collab notebook.
requirements.txt CHANGED
@@ -1,8 +1,10 @@
1
  librosa
2
- numpy
3
  opencv-contrib-python
4
  opencv-python
5
  torch
6
  torchvision
7
  tqdm
 
 
8
  numba
 
1
  librosa
2
+ numpy=1.22
3
  opencv-contrib-python
4
  opencv-python
5
  torch
6
  torchvision
7
  tqdm
8
+ pytest-runner
9
+ paddlespeech
10
  numba