Expanding what Hawk said, you would actually write the story (and include descriptions of scenes), make a storyboard. 

 

The way the pro's do it is to do the voice acting first. That way you can build the CG part of the project to match what is finalized, not finding yourself going off on a wild tangent and having to re-record half your dialog. That is the part that kills a lot of projects; "But wouldn't it be cool if----!!!!!" And the smaller the project, the less tolerance you have, as change after change eats time. And if you pissed off one voice actor with repeated demands, you might find yourself having find or hire a new actor and recording all the affected characters' lines again.

 

I some ways, if you want to get done, you sew your bloody hands and genius to the floor. Finish the script. Script is God. Finish the voice recording. VR is God. Wash, rinse, repeat. You'll never be satisfied, as you can always find things to improve. To tweak. To swap out for the new thing. And you will procrastinate yourself into exhaustion....and out of time.