-
In this assignment, I started off by exploring the Wav2lip repository. I found out about the GAN and simple model. The video that i have provided is made with the GAN model.
-
A problem that I faced while using these models was, they used to move lips even in frames where face was not present. To overvome this I developed an alternative approach with code, I manually extrated the clips in which face was present and in which face wasn't.
-
Used the videos with faces with the model and the other video were simple synced.
-
Unfortunately, the resulting video has blurry frames and glitchy video.
-
I took help from BARD from google for this assignment, also my experience in coding helped devise the second approach.