Jeff Evans
All analog beasts are another story but the fact we have such power now and nearly all aspects of analog circuits can be modeled is rather interesting. The fact that some virtual versions of these things such Arturia devices and Sonic Projects etc clearly shows that to be the case. Dave Smith is even using digital oscillators now in most of his new analog synths because as he says they sound fantastic! Plus more interesting options available too.
The previous generation of DSI synths used Curtis digitally controlled chips as oscillators and filters, the waveform being analogue but the clocking (which controls the frequency) being digital, so none of the tuning issues associated with voltage controlled gear. Which some voltage-control fanatics moaned about, and the Curtis sound is, like any sound, one you get on with or you don't. The chips were originally intended for control circuit use I believe, but also happened to be most of a synth on a chip or two.
Anyway, Curtis discontinued the chips so DSI (and Doepfer, who used them in the mark 1 Dark Energy) presumably had to rethink their approach in a world where discrete components means expensive build costs and the market seems to be much more for good, inexpensive gear than high-priced synths for the well off only. The UK price of the original MS-20 was eye-watering, now the excellent Korg 'reissue' is affordable. And Korg's software M1 fills the same role regarding its ancestor very well. It's even easier to programme, less menu doving and button pushing.
Hardware synths that were simply a dedicated computer in a box with a built in DAC should transfer to computers with at least as good a quality of sound. Modern PCs and Macs have a heck of a lot of power built in compared to ten or fifteen years ago. Waldorf's Nave is software only and excellent, as is their Largo. But neither (nor my digital microQ) have the same feel as a Pulse or Pulse 2. Come to that, Largo doesn't quite sound as "Waldorf" as the microQ now I think about it.
Where emulation does fall down I think is, to pick on Korg as an example, when it tries to emulate something both distinctive and a little non-uniform in its response. Korg's software MS-20s, for computers ot iOS, do not sound like or feel like my MS-20 mini. Close, but not there yet. They do make a handy kind of visual patch library for the real thing though - copy the hardware patch into the software, save as a preset and next time that sound's wanted there's a convenient visual record of the settings. Much easier than remembering to take photos of the synth and make notes in the hope that next year I'll still know what that patch was used for.
Similarly, I just don't find amp/fx emulators comfortable to use. Many people get good sounds out of them, but I prefer hot glass, analogue distortion and analogue filters/wahs. I can't explain it any other way than that. It doesn't help that I can build an "exact" replica of what's been my standard guitar setup for quite some time, every pedal/fx modelled, the amp and speaker models are even "authorised by the amp manufacturer" and..... it's not even remotely close. The core amp sound is very wrong, both in terms of tone and gain structure and the fx likewise. The modelling is also much, much noiser in the way of hum and buzz than the hardware.
Though I'm quite happy to run guitar-compressor-wah-muff-tube screamer-amp-speaker simulator/load box then add modulation, delay etc. in the box. It's that initial sound production that, for me, still needs to be hardware. It's a nuisance actually, having everything ready with instant recall would be very useful.