Not sure how easy this would be to implement, but would be super cool if you could support stems instead of just master stereo bounces. That way you could sync up visual elements with musical elements (percussion, FX, bass etc) with finer control than the freq band ranges. This could create even more immersive/expressive visuals.

Part 2 could leverage stem separation algos to do the above with just a master file.

I think you could probably achieve the above using separate projects for each stem (right?), but having it all in one project would be super convenient.