The system creates your virtual self what is does is make an image come alive. It takes only a close-up image of you and a text input and a gender choice for the voice that you want and once done. The system will create a video of your image speaking the text that you provided and email it to you on your given email id.
Checkpoints can be found under following link: google-drive or yandex-disk. Place the checkpoint in the checkpoint folder
The models were hosted on git LFS. However the demand was so high that I reached the quota for free gitLFS storage. I have moved the models to GoogleDrive. Models can be found google-drive. Place the model file(s) under sda/data/
We support python3. To install the dependencies run:
pip install -r requirements.txt
Run the Final Code.py (Check default parameters) This takes 3 inputs
- Image
- Text (Both should be present in inputs folder. Text has to be edited in the inputs folder)
- Gender (For voice type - male or female)
python FinalCode.py --source_image pathtoimage --voice male/female
