I skimmed the datasheet - it's an interesting and complex device, thanks for pointing it out!
Precision digital long-distance C measurement requires high:
1. absolute C sensitivity
2. voltage swing
3. bandwidth
4. stability
The final bandwidth can be fairly low (I personally wouldn't go below ~50Hz or so) but if environmental noise isn't dealt with at higher sampling rates it can alias in at whatever lower rate the process is operating at. And mains hum filtering is really essential (though you can do this in software if the raw sampling rate is high enough).
Can this device give you high voltage (>20V p-p) swing?
I've already worked all of this out in FPGA logic, and I wish you were closer so I could show it to you in my "lab". May I suggest that we pick a small FPGA board for your development efforts? I could give you a load that would interface with your processor in any way that you like. I could give you SPDIF too. It's all quite plastic so you can do just about anything in fairly fast logic.
At the end of the day, unless manufacturers are specifically targeting Theremins for their products (don't hold your breath!) what they do offer, and what might work after a fashion, is really designed for some other very specific application / industry (tablet / PC, automotive, appliance, etc.). This applies to coils, integrated C / L sensors, etc. Generally (and unfortunately) the best Theremin solutions must still be hand-crafted (IMO). You are already winding your own coils, you might as well take the next DIY step with an FPGA. This would free you up to concentrate on the software side of things (linearity, voice production, user interface, etc.).