There are 2 repositories under visual-to-sound topic.
Source code for "Visually aligned sound generation via sound-producing motion parsing" (Published at Neurocomputing)