You need the VCA. That, in fact, is exactly why they exist. For audio, use an exponential so that the response behavior works in line with what we psychoacoustically expect as a "loudness curve". Or better still, plan ahead by adding something like an Intellijel Quad VCA, which not only gives you four to work with, but tailorable response curves as well, plus mixing capabilities. As your system grows, you WILL need more VCAs, both for audio as well as linear ones that're DC-coupled (which the Quad VCA ones are) for CV/modulation level control.
Also, if the VCO and EG are all you have at present, you're going to need a VCF so that you can use the EG to also vary the timbre. One I'd suggest as a great starter would be TipTop's Forbidden Planet, which is a clone of the Steiner Synthacon multimode VCF...and I can tell you from experience with the actual thing that that's a VCF with character for days. Plus, for what it is, it's quite cost and space-effective.