The late Bob Moog solved this problem in a very interesting and creative way when he built the Ethervox MIDI theremin with software designer Rudi Linhardt.
The E'Vox will operate in five "modes", one of which is PITCH SMART MODE which operates as follows (as quoted from the manual).
"As the left hand pulls away from the volume antenna, the amplitude of the theremin volume control signal increases and is converted into a value that is displayed on the performance screen. At the instant that the left hand's movement is reversed, a MIDI Note On message is sent with a Note On velocity value equal to the last value displayed. The Note On number is determined by the position of the right hand in relation to the pitch antenna."
One of the problems of attempting to play a MIDI keyboard or module using a gestural trigger is that notes produced by pitchbend have quite a different sound from what is produced by the keyboard (or keyboard triggered module) itself. For example, a cello program C2 pitchbent up three octaves is totally different from the sound of the same program at C5, although they are both the same pitch.
Bob Moog told me once that the way around this would be to create a "lattice" of interlaced faders. Each note would be triggered separately in a gestural sweep but the effect would be seamless because as each new note swelled in volume, the preceding note would fade in precise sync with it so that no audible artifacts would be produced. No individual note would have to be pitchbent up or down any more than a halftone.
Sadly, this technology was never explored and we have since lost our dear, wonderful theremin mentor, Bob Moog.
One of the things that distinguishes the RCA theremin (and vacuum tube theremins in general) from their transistorized counterparts is that the timbre morphs slightly from register to register giving the instrument a very "organic" and often uncannily human sound. This has to do with the heating and cooling of filaments as the hand moves within the playing arc.
Interlaced faders, as proposed by Bob Moog, would do the same thing for a MIDI interface. On a keyboard sampler where you have a separate sample for every minor third, the sound would shift in sync with pitch hand gestures.
I don't know a great deal about these things and it is quite possible that these days there are easier roads to Rome.
