I’ve been quite busy going through piles of old scores and deciding what to do with them. I came across a piece I’d written in 1981: Fantasy Music 1 for Flute and Computer. It was a type of piece that was fairly common in those days and involved a tape backing with a live player. The backing was created at the old MIT computer music lab—a predecessor of the Media Lab. Making any sort of computer piece was really quite difficult in those days and required access to a mainframe computer and many hours of computer time. Once finished, the piece would be recorded onto analog tape and played back in the hall with the live performer playing along.
I was very fortunate to have the piece premiered by the excellent flutist Wendy Stern. She played it first at Kresge Hall at MIT, and then at Julliard and around New York. She also recorded the piece for the Composers in Red Sneakers LP. It was picked up by a few other players, but I’m not aware of it having been played in some time.
Some years back, I found a massive printout of the computer data for the piece. It took up a few hundred pages of a dot-matrix printer. This was the way you did backup in those days. You’d get a call from the lab. They needed disk space and were about to delete the piece. So you’d dash over and print the whole thing out. I’d tucked the stack of papers into an fat envelope and thrown it on the bottom of a pile. When I discovered that I still had the printout, I scanned it and ran an OCR (optical character recognition) program to convert the data back into text files. As it turns out, the computer program I’d originally used (then called Music11) still lives as a program called CSound. So I could regenerate the computer part of the piece on my computer here at home. As computer languages go, it’s quite primitive, but there are few languages that can still run programs 40 years old. I thank the group that’s kept it alive.
Back in the 70s and 80s you had to make a number of real sacrifices to generate a piece on computer. Most of those were related to processing time and they had a deleterious effect on the sound. I had to make compromises of that nature to generate this piece. I’d show up in the lab at 8 PM or so, kick off the job and sit there doing something else until the job finished at 6 AM or so. It took all night—with the computer all to myself—to generate just a few minutes of music. It took many of those nights to get a piece just right. Most computer pieces of that era have a hallmark sound. I think that’s the result of it taking too long to really explore and perfect digital instruments. Although the language had potential, it just took too long to experiment.
So what to do with this particular piece? I think there are still plenty of recitals that use music like this. So I decided it was worth it to at least generate a clean backing track. I eliminated those compromises in the code and adjusted the areas that caused distortion on the tape. But I also decided to keep the basic sounds as they are—just with much lower noise. It’s a piece of its time, and I should respect that. Besides, I didn’t want to spend any more time in CSound than I had to! I also realized I could put all the instruments onto separate audio channels for better mixing and to open up the possibility of an Atmos mix at some point. So instead of the basic stereo output, I generated a 40-track output and tossed it over to ProTools for mixing. And just to show that computers have progressed in all those decades, here’s how long it takes now to generate that clean 40-track output: 12 seconds. That’s right. A job that used to require 9 or 10 hours now takes 12 seconds.
If you’re an interested performer, you’ll now find that you can request the music and the backing track (now just a wav file for your laptop) right here at no charge. It’s been quite interesting to revisit a 40-year old mindset and some ancient tools. It’s also fun to freshen up an old piece with some challenges and listener appeal.