In animation, it takes several hours to create only a couple seconds of visuals. As an animator myself, I often become overworked with the harder smaller parts of animation that just end up costing more time such as lip syncing an animated character. These are the visemes I created and used to animate lip syncing.

In my project, rather than using audio, I decided to focus on the video of my mouth speaking. To do this, I wrote software to locate where my mouth was in frame then passed it through a machine learning model to predict what mouth shape I was making. Overall, the computer does only a fair job at creating a convincing lip sync. (The bar graph shows how confident the computer was at it's answer. The red bar denotes which one the computer picked):