Got hired by a theater as Master Electrician and decided I needed to learn more about sound engineering. Mixing, in particular. And for no good reason, I started composing things again just so I'd have some multi-track material to learn mixing on.
It didn't actually help with the mixing. I learned that elsewhere. But it got me started on including original music in my theater sound designs.
More below the fold because bandwidth.
The first big tool was a Roland W30 workstation. Which is to say; a 76-key sampler -- with a large library of patches on floppy disk -- with an included mini-sequencer accessed through the small LCD.
I was very happy when I finally managed to find some sequencer software that would run on my ancient computer (I've almost never owned a computer less than ten years old). Then came "Lend me a Tenor"; a design requiring orchestral backing tracks, and the W30 just didn't have the polyphony.
So I added a Roland M-OC1; basically, a ROM expansion card from their then-flagship synthesizer in it's own stand-alone rack-mount box. I followed that up over the next several years with several other modules; Roland M-SE1, D-110, and a JV-880 with the "Pop" expansion board.
A snippet from the final cue, with our actors mixed in
When the system was up and working I could quickly find the sounds I wanted from the sequencer -- full pop-up lists of all the available modules and their patches -- and I had polyphony to spare. I was able to do things where I broke down a "chord" into different (and active) individual lines for each violin section, violas, 'cellos, etc. I even had a template that presented the usual resources of a small symphony orchestra in symphonic score order (aka flute on top, piano just above the percussion, etc.)
A sketch done as a loading screen for Ambrosia's "Escape Velocity" game
But of course every time I migrated to a newer computer, or added gear, or took things apart to move or clean, it didn't quite get back in the same order. So every song I was still tinkering with would become an artifact of a previous layout; the specific sounds that made that arrangement work were either somewhere different or didn't sound anywhere near the same.
I remember using one distorted electric guitar patch from the W30 libraries (aka the Roland S550 and S330 libraries, on multiple boxes of floppy disk) for a wild little guitar solo. And when I migrated to using mostly JV-880 patches instead of putting up with loading floppy disks every time, it never quite sounded right. So my files are now littered with half-completed sketches for which the MIDI files now are useless; only an audio record remains.
Eventually, though, it got to be too much. I needed space, and the number of cables to hold all those rack units together was too messy and high maintenance. I played around with hosted DSP; I managed to score a nubus SampleCell card, and a 16-channel sound card, which both went in my first frankenmac (a nubus machine with a G3 daughterboard). And I never actually used them, more's the pity.
Instead I went to computers fast enough (barely) to handle everything on their own hard disk and CPU. The first purchase there was probably the shareware VSamp Pro, which I am still using. And the first library was Garritan Personal Orchestra (followed in a few years by Jazz and Big Band from the same company, and just last week, World Instruments).
But, oddly, I haven't gotten a lot of use out of those orchestral libraries. The one show I did with a lot of original material, I used almost entirely freeware and shareware software synthesizers, and patches I created myself out of industrial samples and so forth on VSamp Pro.
See, one of the reasons to go software is that the framework of a VST host -- and modern DAWs -- allows every individual instrument to be processed individually. Instead of writing things with a full string section parted out, two f. horns with their own individual lines, parts for Bb and bass clarinet; for that show I wrote with piano, pad, and effects. Or alpenhorn, solo 'cello, and found object percussion.
The different approach being, the sonic space was filled not by multiple lines from symphonic instruments with different timbres, but with flanging, echo, reverb, distortion, etc. So there were fewer instruments playing, but each had effectively a thicker texture.
And of course each individual channel -- whether carrying a MIDI instrument or a recorded line I played myself acoustically -- had multiple available layers of equalization, compression, and other processing.
The opening cue from a staged reading of "Agamemnon."
We used a different translation in the show.