Start with synthesis. I've been listening to both recorded and live symphonic music, and synthesized versions of same. And I think I've modified a previous opinion. There's something intrinsic to the "real" orchestra that gives it that power. An intangible something, not easy even for a sound designer like me to try to point to.
At first glance synthesized orchestral instruments would seem to have it over synthesis of solo acoustic instruments in being convincing. And at this first glance, yes. A synthetically produced backing track can natter on in sub-John Williams for hours without attracting attention to its nature. But then you put the real deal in your ears, and there's an excitement and a power and a presence. Somehow, all those little bits of noise and blurring and sweat and spit that mark a mass of real humans all working in concert comes through: and the effect is inspiring.
And, oddly, synthesized solo instruments can often fake it more convincingly. Well, some. Plucked strings like guitar fare better than wind instruments. The extremely vocal nature of the saxophone makes the real deal stand out. Oddly enough, there's something about all the brass that doesn't take to synthesis. It always feels somehow off. Paradoxically, I find the synthesized brass that is most emotionally stirring is that which most demonstrates its artificiality; the thick "brass" sound of DX and JV synthesizers as heard in disco and 80's pop. It is as if there is an uncanny valley; "almost human" sounds worse than "clearly artificial." But on the flip side, a synthesized guitar can leap out and grab you with that intangible "reality." And this isn't even, oddly enough, a result of modern synth patches being built on full-length note samples played on actual instruments, because the same prickle of hair on the back of the neck can be raised by a guitar sound originating entirely in physical modeling synthesis.
Well, on the practical side, putting the humanity into synthesized tracks is work. A lot of work. Garritan's libraries introduced some clever tricks to allow a keyboard player more of the expression natural to a wind player (at the cost, of course, of even more concentration necessary to record the track in, and more time spent in laborious hand editing when the take didn't quite go right). This may be as simple as throwing in a few guitar lift-off and fret squeak noises (which go a surprisingly long way in fostering the illusion), or it may be as laborious as recording each and every violin in a section individually (a tremendous amount of work but the results are startling).
When I was doing my own pop-orchestral synthesis projects, I struck a compromise by breaking down to each desk or chair; recording two or three times for each section of violins, then again for violas, 'cellos, basses. Besides giving a better sound than a "string" patch, I think the internal movement possible when doing this is a heck of a lot more idiomatic to that massive complicated instrument we call the symphony orchestra.
If you simply must have exposed solo lines, then there's an old bag of synthesist tricks. Add little bumps to volume and (even more subtly) pitch. Hand-add your vibrato rather than trusting the patch programming to handle it. For that matter -- I used to write string lines with hand-fingered tremolo. Which is pretty much the same reason that playing in a track on keyboard (or other MIDI instrument) is superior to dropping it in mechanically with editing tools.
None of this is new. I realized way back on my second sound design using orchestral synthesis that I favored a performance by even an amateur human over the sterility of the synthesized material. The best of both worlds being using a little of both (as was recognized very early on, to the extent that a major package for synthesized backing tracks for musical theater cautions that the more parts you replace with live players, the better the result will sound).
Most of the game music covers I have been admiring of late showcase a soloist on a real instrument, seated against backing tracks that are better able to hide their synthetic origins. Of course one has to have a musician -- better yet, a singer -- capable of doing justice to the material. Because on the far side of that Uncanny Valley is the perception of a real human who is playing wincingly out of tune, and that can be even more distracting than a clearly but unabashedly artificial player.
And thus we come to my own somewhat prickly relationship to music. I'm not a musician. I lack a lot of basic skills. More than that, I lack the "heart." There's something I can hear in everything I do which is a lack of soul. So perhaps I am best suited to exactly that sort of solo MIDI composing I seem to be disparaging above. Except even that fails if done mechanistically. You can be precise, you can use the technology, but you still have to have those sensitive artistic choices.
Sure, I lack sufficient interest to gain the other basic skills. I think I have pitch sense. I can stumble around in sheet music but can't really read. Certainly not sight-reading. And my memory is horrible. Working in musical theater as I do, I am surrounded by people who can remember every note in a long song. I get lost if a tune is any longer than the Westminster Chimes. And this would probably have come with practice. I remember starting up at bouldering walls once wondering if I'd ever be able to memorize an entire sequence. Which I do so easily now it isn't even worth remarking.
Oh, yes, and at some point so long ago I can't even remember it I must have sat down and learned my scales, because I can still go through all the majors without having to think about it (the minors come a little harder).
I'm not sure which I hate worse; the things I am conscious I lack the internal wiring to ever do well, or the things I hold the (possibly mistaken) opinion that I could learn to do if I only had the time.
I do know that the vastly larger portion of the time I have spent working with computer music has been technical labor. Organizing patch libraries, editing samples, plugging and unplugging gear; constantly trying to come up with a rig that I can just sit down at and play instead of having to fuss with every time. But the Red Queen's Race of technology can not be won in this arena either. As fast as I learned one piece of hardware or software, a better one would come to replace it. In the end all I had to show was tinkering, and a few scraps of pieces created on equipment so long-gone it is useless to think of continuing work on them.
And all in all, I would have done better -- I still might do better -- to put aside the computers and spend a little more time with my ukulele. (Which, oddly, is the only musical instrument that gets any attention these days. Dust is collecting on my new Behringer controller keyboard, and I've almost totally forgotten how to work in Reaper, but my uke sees almost daily use. A strum here and there is so very relaxing.)
I've been tempted for quite a while now to try to combine the opening tracks to classic Tomb Raider and the series Stargate SG1. And for that matter tinker up a jazz interpretation of the Black Mesa theme. But to close for the moment, here's the last full-length piece I did for my own amusement. Which was more years ago than I'm comfortable thinking about.
(And here's the fiddly little details. This actually came out of that laborious and possibly pointless process spoken of earlier; collecting and editing patches. Found a freeware "old music box" patch and as soon as I had it connected to a keyboard started playing a sort of Danny Elfman inspired ostinato with intentionally quirky chromatic development. I can't say "harmonic," because I am still grossly ignorant of that whole aspect of music theory. Anyhow, built on it with mostly Garritan patches, and added a few layers of sound effects. The closest this whole thing gets to the idea of incorporating real performance is that all the foley is mine. I walked a pair of mary janes across a wooden floor on my hands, for instance. And then did a whole bunch of manipulation on those samples, of course....)
(And that's actually another idea I've been tinkering with, ever since seeing a video on how a certain mechanical music machine was recorded. And that is to make use of available acoustic instruments -- cheap ones or improvised ones -- with the intent of capturing the human element of the performance and some of the noise and grit that gives it a grounding in reality. But then processing the audio to make it sound better. The kind of dial-tweaking to bring out the essential character that I've been doing on sound effects for years, really. And come to think, I did this once; my work for a children's production of Mulan included my processed version of a handful of dowels I rapped on the floor as a percussion element.)