/Wav2lip_SuperResolution

Wav2Lip with ERSGAN

Primary LanguageJupyter NotebookMIT LicenseMIT

Wav2lip_SuperResolution

README

This repository contains a Wav2Lip model with GAN for improved lip syncing and employing ESRGAN for High Resoultion video of LipSyncs.

To use the model in Google Colab, simply click on the following link:

https://colab.research.google.com/drive/1F6Bj8G1eiktudzuusKYdHRiJOPlpA4-L?usp=sharing

This will open a Google Colab notebook with the Wav2Lip model.

Sample Inputs

Video-https://openinapp.co/5cwva

Audio-https://openinapp.co/o9vuj

Usage

To generate a lip-syncing video, simply follow these steps:

  1. Open the google colab with the link provided above.
  2. Install the pre-required dependencies and the Wav2Lip model.
  3. In the Google Colab notebook, mount your Google Drive account.
  4. Upload the video file that you want to generate a lip-syncing video.
  5. Upload the audio file that you want to generate a lip-syncing video.
  6. Then in last cell to churn the data add specifiactions of padding based on your own needs.
  7. Then finally run the last cell to generate the lip-syncing video.
  8. Finally connect the Lipsynced video with ESRGAN part install the pre-required dependencies.
  9. Train on each frame.
  10. Get Final Video.

Sample Output

https://drive.google.com/file/d/1uNVv3ncr8J6YutaG02OcGlV9K3nSATV8/view?usp=sharing

Contact

If you have any questions, please feel free to contact me at truelykush18@gmail.com.

License

This repository is licensed under the MIT License.