Many videos I've watched on synthesis aim at emulating physical instrument sounds (or sometimes physical sound effects), with the pursuit of realism as a guiding principle. At first, this struck me as something in the neighborhood of a skeuomorph. I wondered why the contingent attributes of physical objects should dictate sound design principles now that their limitations can be put aside using synthesizers, whose space of possible sound outputs is much larger (at least if we ignore differences between indiscernibly similar sounds).
Later I discovered the answer that our ears have evolved to appreciate the sounds physical instruments and objects can produce, because they have been with us for so long, but I still wonder how musically useful synthesis can be for producing original sounds rather than emulation.
Very qualitatively, what percentage of possible sounds a synthesizer can output are viable for music? I know an exact numerical percentage would depend on way too many factors and boundaries, but I just want a general sense of how sparsely musically-viable sounds occupy the space of all producable sounds.
Again very qualitatively, what percentage of those musically-viable sounds correspond to sounds produced by physical instruments or objects? Is that evolutionary correlation between the ear and the physical so strong that it accounts for nearly all of our ability to find pleasure in musically viable sounds, or does synthesis unlock a large amount of musical potential not grounded in emulation of the physical?
If synthesis does open these original avenues, is there a field of study which teaches how to create music by designing these original sounds, rather than by emulating physical instruments or objects?