NOTE: This page is out-of-date. I made a new homepage in January, 2007 (to replace this page's dead successor). The page you are looking at has information not replicated on the new site. However, this material is now mostly of historical interest.

First of all

I will take preemptitive action. I will answer your question before you e-mail me to ask it. "Experiments in Musical Stupidity" is a title (like much of the rest of my site) alluding to obscure external sources. It is a modification of the title of a project by David Cope, "Experiments in Musical Intelligence", intended to immediately convey to the informed reader what type of work is presented on this page.

Computer-generated music

Since you already have a fair idea of what this page is about, I will dip you straight into the details.

There are two approaches to getting a computer to compose music. One is purely algorithmic: find a cute way of generating note values and their duration and put it to work. The other is large-scale intelligent pattern matching: find commonly occuring note-duration patterns in some set of musical pieces and the commonly occuring patterns in which those patterns occur (and so on for n levels) and use these data to put together something that sounds very much like the given set of musical pieces. You say: "What?"? Well, I meant to write that the other approach is imitating existing music, but that would sound silly, would it not? The two approaches can, of course, be combined.

The second approach is by far the harder to implement: it takes much parsing and format conversion, and, in addition to that, good knowledge of musical theory and of the that music one is trying to imitate. The results of the second approach are boring. You've heard it all before.

As can be easily seen from my rhetoric, I favor the first approach - algorithmic. My reasons, previously intoned, are that I want originality, I know almost no music theory, and I hate parsing anything but raw numbers in the format I want them. Besides, David Cope already brilliantly implemented the other method. You can search for "Experiments in Musical Intelligence" and will certainly find great reviews and pleasantly sounding samples. Since the second way is difficult and thought-intensive, my meddling could not, in comparison, be titled "intelligence". That left me with no choice but the direct opposite - "stupidity". As you will see from the samples, the choice of name was perfect.

I took a pink(1/f) noise source for each of the voices in the piece I wanted to generate. I introduced repetition into the piece by making one voice echo the other with some delay. I forced the notes in the different voices to be consonant. I wrote a decent synthesizer. As the versions of the program progressed, I also twiddled with rhythm and accompaniment a bit. Then, I gave the project up for reasons that shall soon be transparent.

Two months have passed. I remembered about the project. I listened to some of the mp3s that survived. I was not motivated to improve my horrible source code. I thought that the world should know. I found all of the remaining pieces, re-mixed those that I could with an improved synthesizer, named them, filled in the ID3 tags, and put them on the web together with this explanation. I also realized that perhaps my synthesizer isn't too bad and could be helpful to another poor soul tortured by the demon of [artificial] music. I tidied up the source code, wrote up the documentation, and away through my modem the synthesizer went.

Now I will prevent another one of your questions. What kind of an album name is "Surviving Experiments"? A bad one, I admit, but still that is what those things below are. Products of twisted manipulations of a maniac. Pure, innocent pink noise contorted into unthinkable horrors.

This ends my confession. I leave it to you to judge me. Have I done humanity a service by proving that creativity is impossible to simulate, or have I destroyed life's meaning by making a machine create art? Feel free to deposit this question into the minds of all of your friends and aquaintances together with the works in question (in legal terms this means that you are granted permission to copy, store, and distribute the mp3 files below in their unmodified form, free of charge only; a separate license applies to the synthesizer).