Visine options in 'compute lip sync from scene audio'

“Surprised” is actually a built in viseme, as is “smile”. The supported visemes are in the tags panel. I am guessing you don’t have it tagged in the puppet. But if you want to go beyond those (e.g. “angry”, “sad”, “sleepy”) you need to use a different approach with triggers.

My recommendation is to create “Mouth Group” under which you put “Mouth” under which goes “Ah”, “Uh”, ... “Smile” - the standard visemes. Then you create additional layers under ”Mouth Group” (siblings of Mouth) for all the non-standard visemes. You then create a swap set with “Mouth” being the default, and “Angry”, “Sad” etc triggered by a keyboard triggers. So the default mouth comes up for lip sync, but you can override it with an expression with a keyboard trigger.

It‘s not perfect, but its the best option I know of.