Nah, not 64 voices. Let's say you have an FH-2 plus a full compliment of seven expanders, so 64 outputs. That means you can break all of this out into, say, eight monophonic channels...with each channel having:
1) note CV
2) gate
3) velocity CV
4) aftertouch CV (poly-capable!)
5) assigned modulation controller CV
6) release velocity CV (yep!)
7 and 8) two more sets of eight for special modulation purposes, assigned as needed.
Now, for a live gig situation, this is a bit over the top, but you'd still want the first four. But when using the FH-2 via a MIDI interface, you can have your DAW send all of that stuff and the FH-2 will break it down as needed. And for complex generative work, having massive incoming control signals to up the complexity ante is always a plus. Plus, having the X and Y inputs in use as a pair of 'sense' circuits, able to send CV back to the DAW, means that you can continue the generative architecture on into the computer itself; something such as Ableton Live + MAX for Live would allow you to build up specific M4L objects to listen to and react to those return 'sense' lines for the M4L object(s) purposes. That would allow a generative process with full control feedback between modular and computer, and endless control-order possibilities!