I wouldn't add more voices. More modulation and VCAs, probably. Maybe something like a WORNG Vector Space or something like that where it makes a lot of modulation out of a couple sources. A matrix mixer is good for that, too - you can mix different CV and audio sources to make unique shapes and adjust them on the fly. It seems pretty complete to me, otherwise.

Was there something specific you wanted it to do? It might actually help if you're specific about which parts work less for you.