Apologies if this thread seems fairly opinionated, without any audio samples to back these claims up.
So–I've finally had a chance to tinker around with my DSI Pro-2 + Tom Oberheim MIDI-to-CV SEM combination, mostly in an attempt to offload patches that I would normally use the SEM for, to the Pro-2 for ease of recall, etc. It's hardly scratching the surface, mind you, of what the combination can do, with or without CV patching into the SEM from the Pro-2–that said, I think I've got a handle on some comparisons between the two voices:
- The Tom Oberheim SEM filter response seems to vary with oscillator pitch, even when keyboard tracking is turned off, making it quite complicated to match the filter response between it and the Pro-2's SEM implementation (which does not seem to vary when keyboard tracking is turned off). Of course, the same also applies once keyboard tracking is turned on–and as the SEM filter does not self-oscillate, it becomes quite difficult to guess where the variance occurs; I believe that I had to add some additional negative tracking to approximate the response at the Pro-2.
- The SEM ramp/sawtooth wave contains more fundamental than the Pro-2 sawtooth, so it is necessary to add an additional sine wave underneath to match the timbre. The girth, drive or air controls do not change this.
- Once you have a reasonable approximation, the Pro-2's girth ( 64 ), air ( 32 ) and drive ( 8 ) controls can be used to mimic the even-harmonic distortion that occurs when the Tom Oberheim SEM filter resonance is in the uppermost position.
- The two units together, panned slightly within the stereo field, give you a pretty good idea as to what a Two-Voice (Pro or otherwise) might sound like. And it's awesome.
So my shallow recommendation is that the Pro-2 + SEM combo is absolutely killer–but you might have guessed that anyway.
Also - it is most definitely possible to approximate the sound of a monophonic SEM voice on the Pro-2, provided that you spend a little bit of time to add some chaos to your final patch.
I'm gonna guess that the OB-6 variant of the SEM design likely has compensated for many of these irregularities in response, some of which would be quite beneficial, others which might very well be somewhat annoying if you were trying to match the response across multiple SEM units (or SEM filter-equipped voices).