In Pictures of Sound, I defended my efforts to educe audio from handwritten medieval documents as though they were sound recordings by arguing that many modern-day “sound recordings” aren’t records of past sonic realities either. I owe that perspective partly to the experience of making recordings of my own MIDI compositions between 1997 and 2004, such as the friendly little earworm “Lyrics About Love” (June 2003):
In spite of the title, there aren’t actually any other lyrics; I just imagined a refrain beginning and ending with the phrase “lyrics about love” and left it at that. At one point an online greeting card company ran across the MIDI online and contacted me to ask about licensing it. I’m curious what kind of card they had in mind to pair it with, but somehow I never got around to replying, so we’ll never know.
I haven’t thought much about my erstwhile dabbling in MIDI for the past ten years. After all, I saw it strictly as a hobby, wholly unconnected to my research on historic sound media, and I’d set it aside when I started writing my dissertation, although I never took the old MIDI page down from my website. In retrospect, though, I wonder whether these experiences could have shaped my thoughts about recorded sound more than I realized. It was in roughly those same years—1997 through 2004—that I began collecting 78s and cylinders in a serious way and studying the history of sound media and performance as a grad student in Folklore and Ethnomusicology. And it so happens that nearly all the “sound recordings” I was then making myself were synthetic ones, created using computers and sound cards with no human performers and no original sound waves passing through the air. Maybe that helped nudge me towards imagining a similar category of “historical audio” years later: one based not on the recording of sounds out of the air but on representing sound in particular ways.
It was back in 1997 that I stumbled upon a piece of software called NoteWorthy Composer (NWC for short). Up until then, I’d enjoyed composing music but had been frustrated at having no convenient way to actualize any of it, or at least anything that was written for more than one instrument or that exceeded my own limited ability as a performer (I play piano and accordion, but mostly by ear). NWC promised to change that. I didn’t care so much about being able to create staff notation—I could do that by hand on paper—but I was excited about the prospect of using the MIDI features of the software to play and hear things I’d composed at the click of a mouse.
I started out by keying in a couple pieces I’d already written, but the following was my first real effort to use the software to compose something from scratch, and an unabashedly kitschy piece of work it is too. I give you “Squidbop” (1997):
I had to enter individual notes using staff notation, but I also hit upon a few strategic efficiencies that I exploited shamelessly. I could program in a sequence of bars and copy and paste it a few times; then I could vary things by adding or subtracting other lines in counterpoint with it, or repeat it assigned to different voices in turn, or auto-transpose it up or down a note or a whole octave. “Squidbop” already featured a few of these strategies, but the next piece—“Out and About” (October 1997)—is another typical example of what I came up with during my first few months of playing around with NWC:
I should stress that I was composing these pieces specifically for MIDI, and not just using MIDI as a means of playing them. I liked the sounds of particular synthetic voices and the way certain voices sounded when combined together, and I paid as much attention to that factor as anything else. In fact, when I compared my own MIDI files to the ones I found online, my main impression was that other people weren’t taking full advantage of the available sonic palette—nobody else’s files sounded as timbrally rich to me as mine did, although they outdid mine in other ways, and of course everything ultimately sounded a bit like video-game music anyway. It hadn’t occurred to me that the timbres I associated with the voices I was using were actually specific to the model of sound card I had in my computer. A banjo is a banjo and an accordion is an accordion, right? Well, wrong—but I didn’t know that yet.In the late nineties, I was still relying on a dial-up modem connection for home internet access, like most other people I knew. File sizes that seem trivial today were then prohibitively large. Under the circumstances, MIDI was the only file type I thought was suitable for presenting music on the web, and I embedded my own MIDI files in some of the first websites I put together to give them some multimedia oomph. I was also active at the time in free-form roleplay (FFRP) via text-based Internet Relay Chat, and specifically in the #BlkDragon*Inn community with its own self-created fantasy setting of Arangoth. Many of the MIDI files I posted online were intended to function as music for the fictional cultures within that setting, including this Khalar Dance (August 1998):
At one point, I recorded a bunch of my MIDIs onto audiocassette, partly because I didn’t have any way to automate playlists of them on my computer, and partly because I wanted to be able to play them on better audio equipment than my computer speakers. Then, in 1999, I bought my first CD burner. This was installed in a different computer than the one I’d been using to make MIDIs—one with a larger hard drive that also made it feasible for me to create and store WAV files for the first time. The first WAVs I made from my MIDIs were captured through an audio cable connecting the output of the old computer to the input of the new one with the CD burner in it. Here’s a piece I composed in October 1999 with the filename hunger.mid, right about the time I started doing this:
Note the percussive bottle-blow voice in particular. A little while later, I figured out that I could play MIDIs on the same sound card I was using to record them—the SoundBlaster in the new computer. But I also discovered that the MIDIs sounded a lot different when played using the other card. The percussive bottle-blow voice didn’t sound percussive at all, instruments I’d liked before didn’t sound like much any more, and instruments I hadn’t used much before now sounded really good. So I ended up “remixing” some of my older MIDIs to fit the peculiarities of the new sound card, including a piece called “Like Twin Sapphires Were Her Eyes,” a.k.a. Griffon Dance #7. Here it is as originally composed in 1997:
Soon afterward, I finally found a roundabout way to create dedicated percussion tracks in NWC, as opposed to relying on individual drum voices. My first experiment along these lines was “Zotika” (April 2000):
I also converted some of my older pieces to take advantage of the new-style percussion, including one called “I Get Butterflies (In My Stomach).” This was conceived loosely as a satirical response to the song Butterfly Kisses, which I hadn’t actually heard at the time I wrote it—still haven’t—but which I’d heard was a sappily sentimental song written by a father for his daughter’s sixteenth birthday. “I Get Butterflies” was supposed to be the daughter’s reply. There were lyrics too, but perhaps fortunately, I can’t seem to find what I did with them. Anyhow, here’s the original 1997 version:
The differences between the two sound cards had made it painfully clear that I had less control over what people heard when they listened to my MIDI files online than I’d supposed. I ended up separating the MIDIs on my website into two pages, one page optimized for the first sound card, and one optimized for the second one—but of course nobody without those specific sound cards would hear quite what I meant them to hear, and there wasn’t much I could do about that. Well, there was one thing: I created an artist page at the wonderful old mp3.com and uploaded mp3s made from some of my MIDIs there, in the “electronica” category, actualized as audio in just the way I meant them to be. However, mp3.com folded rather abruptly in 2003, and feeling burned by the experience of losing all the setup work I’d invested in it, I never made any effort to recreate the artist page elsewhere.
In 2004, I migrated to yet another computer with another sound card with another distinctive set of MIDI voices that made everything I’d written to date sound wrong again (although it was better for other uses, such as digitizing 78s and cylinders). The audiocassette, audio CD, and WAV versions of my MIDIs now provided my only means of hearing the stuff I’d written over the past few years as I’d meant it to sound, while the MIDIs themselves had fallen victim to digital obsolescence as far as I was concerned. Given this background, I guess it’s no wonder I later came to think of “sound recordings” or “phonograms” more fundamentally as a way to represent sound in specific detail than as a way to capture sounds automatically out of the air for playback.