as to whether it matters if a user can hear it "in a mix" or not - well, great music can be made on a toy guitar or a cheap accordion or, hell, a rubber chicken. But in many cases it's considered 'better' when performed on a more refined instrument. And that's a big part of why it matters at all.
So I got into synths about 5 years ago. The first few years, I researched and tried to get as many of the sounds or instruments that my heroes did - guys who had means and were buying up cheap analog gear when digital hit the scene. But even after getting all the gear I have, I still wasn't making music like they did. I had more to learn. So I *really* studied the actual recordings - what were the sounds, what did the song really depend on, what was essential and made the tracks, and what could be changed in a remix or remake without fundamentally altering the impact of the music, etc. And then only recently, I came across some live recordings from these bands in the mid-80s, and knowing what I know now, I saw just how unimportant most of the gear really was. The sound was coming from the drum production, samples, vocals, effects, the mix, a hundred other things than just like, a Jupiter or Prophet 5 or Oberheim or whatever. In fact, the actual synth patches are pretty straight forward; Pro One+Moog basses, string sounds from Ensoniqs, Rolands, string and percussive sounds from DX7s etc. And so while I personally DO have sensitive hearing, and habitually do well on blind A/B tests, and am very much inspired by rich and dynamic timbres, the MUSIC that I enjoy tends to use the studio more as the instrument than any one synth or whatever. So this whole thing about sound quality had just become a white whale for me, when in reality it makes almost zero difference for what I want to do.
Does this mean I think software is equal? No, but mostly because most software is not designed as an *instrument*, and it lives on a computer, which tends to automatically disengage the connection to my subconscious mind that a real instrument can pull out of me. Maybe it works for other people, but hardware is more inspiring to me. But as for the difference in sound... it's almost never going to translate to a mix, unless it's a minimal arrangement with that specific instrument in razor focus, doing howling filter sweeps or something. And this is where DSI and Sequential have always stood out, even beyond sound and build quality - they don't make synthesizers, they make MUSICAL instruments. And I have yet to find a piece of integrated hardware and software (I invested heavily in Native Instruments and Arturia because I though NKS was the answer) that replicates the same level of joy and productivity that I get from a Dave Smith piece.
But of course all methods are valid, it's just down to what works for you and what your desired results are.