[Joachim Holler]

Not exactly sure about your setup, but you could try putting a compositing tag onto your cube. In the “Tag” settings, untick “Seen by GI”.

Should the outside of the cube be lit?

[Joachim Holler]

You are right, Ben. At the time you asked this, there was no Script whatsoever that tackled this problem. To be honest, Auto Lip-Sync does hardly more than the “Convert Audio to Keyframes” command.

But finally, there is. A script that animates a chosen number of mouth/face images in sync with the specific audio file, just like you would manually. It’s called lipsyncr (, I am one of the developers). It’s actually funny, because we were especially mindful of exactly the things you mentioned. Of course, there’s a free trial and lots of information available:
https://www.aescripts.com/lipsyncr

I really hope this is what you are looking for!

[Joachim Holler]

I know this one is a little old, but,

yes, finally there is! It’s called ‘lipsyncr’ and it does exactly what you described. (I am one of the developers.) In contrast to the other script mentioned, it does actual speech analysis to create viseme–based animations, just like you would with the manual method.

I don’t want it to look like I am mounting the cow for advertising, so if you are interested, you will find all the infos you need on the web!