What Our World Sounds Like Now

JSR Makes Some AI Music

I am in Los Angeles, on top-secret business, for the next few days. I don’t have a great deal of free time, but if you are an LA-based Hinternet reader and would like to get together for a spirulina smoothie somewhere in the vicinity of Basil Rathbone’s sidewalk star, please do drop me a line. I apologize in advance if my schedule is already full by the time you write — it’s a whirlwind trip, but I am trying hard not to be my usual reclusive self and at least to attempt to establish new human connections as I move through our big world.

Also, please let me remind you that the Hinternet Essay Prize Contest is fast approaching its deadline! Get those finishing touches added to your submission and send it in very soon!

In September, 2021, as regular readers will already know, I hit my head in a scooter accident. The impact shook a few things loose, and while no measurable damage appeared in the MRI, I have suspected ever since that it changed me in both subtle and profound ways. For one thing, in my previous life I had been content simply to enjoy music — to judge it, to curate it, to learn about its creators, and sometimes fully to feel it. Post-accident, I find that enjoying music is not enough. Music, I find, is now coming out of me, like sweat out of my pores, and I have no choice but to let it come out — i.e., to “make” it.

One obstacle here is that I am not a musician. Or am I? In the Categories Aristotle speaks of the musical man who is musical even when he is not making music — unlike the sitting man, who ceases to be a sitting man the instant he stands up. Could I have been a musical man for all those years, without knowing it? Are these songs seeping out of me, now, hints of a latent internal capacity that only had to be awakened? Is this capacity latent in all of us, qua human beings, while only some of us are fortunate enough to have it jostled out of us by one or several little traumas?

To be clear, I am not talking about talent — as I am about to show you, I have none. I am talking, in the first instance, about a capacity, something comparable, perhaps, to the capacity to perceive mental images; and in the second instance I am talking about a desire, comparable to the desire to take a pencil in hand and to draw the images one sees. I find the desire more surprising than the capacity — I often feel, these days, like the folks in Close Encounters of the Third Kind (1977) who find themselves instinctively sculpting clay models of the site to which the aliens are calling them. I mean, I’ve just got to make this stuff.

The Three Estates

The music churns up in me willy-nilly: the melodies, the lyrics. The difficult part comes in trying to translate what I am hearing in there, to get it out of me and into the world, a process for which I have no training at all, and, I have found, no real natural aptitude either. Europeans, who all seem to have received proper musical educations earlier in their lives, have often told me that to be able to discern a note means automatically to be able to reproduce it. But that’s just not how this American experiences music. I can’t sing at all, and I can just barely play the guitar or the piano to pick out melodies. Sometimes, if I have a melody in my head, I will press piano keys one by one, or at random, until I happen to find the note I was hearing. I have to count from middle C in order then to learn what note it actually is.

I also have lyrics for the melodies, which usually appear spontaneously in my head around the moment of the first appearance of the melody, though sometimes I write them first and then only later do I find the music to put them to. Writing lyrics is not like writing poetry, no matter how much the vapid poptimists are eager these days to analyze some mediocre Taylor Swift couplet as if it were Edna St. Vincent Millay. When it comes to song lyrics, at least the kind that pop up in my head, a fair amount of cliché is welcome — pure banalities like “Say you’ll be mine” and even “Ooh baby” that aren’t so much language in their own right as recycled fragments of the human cultural history that these songs in my head are in some deep sense “about”. Again, with due credit to David Lynch — in the 21st century, you do not issue the hortation to “do the Loco-Motion” out of any real expectation to see your friends doing the Loco-Motion. You hortate because it helps to move you into a mode of aesthetic receptivity that is at the same time a hard reckoning with the artistic legacies of post-war mass entertainment.

I have been obsessed recently with the Cuban classical guitarist and composer Leo Brouwer — a musical man if there ever was one. I must have watched this performance a hundred times by now, and each time I have detected some new subtlety to astound me and to make me wish to watch again. (I also love his look, and the Cuban revolutionary mural behind him.) Brouwer went on to a prolific career as a film composer, and in several interviews has presented himself as an absolute master of his craft, working according to the highest standards and the most stringent idea of what his compositions ought to be doing. One by one he dismisses as unnecessary, as frivolous, as cliché, the most familiar features of a musical work: rhythm, melody, dramaturgy — you don’t need any of it! What is left over at the end of this process —his film scores—, does not sound particularly lush or lively, but the artist keeps his integrity intact. This is not quite Adorno telling us to eat our vegetables, that modern music, if it is to succeed at what it is trying to do, must sound unpleasant; it is not quite that, but pretty close.

“The general public recognizes [certain chords] as something nice”, Brouwer reflects in one interview. But “the abuse of something nice becomes a banality.” Perhaps, yet it seems to me that in pop music, perhaps somewhat as in the joke drawn out so long that it becomes unfunny and then funny again, to continue to abuse certain well-chosen banalities can sometimes bring them back again from the ridiculous to the sublime. Even “Ooh baby” can never be lost forever.

So I admit that the music in my head is banal, cliché-ridden, worthy of scorn from Leo Brouwer. I also think it’s beautiful. Because it’s beautiful, I struggle to get it out, and what comes out is only ever a dim after-echo of what I had originally heard. On a few occasions I have managed to get out something that at least sounds, melodically, like the internal original —the work itself?—, simply by pecking patiently away at the keyboard or the fretboard. But when I have not been able to do that, I have, recently, found myself turning to technological aids to bring to life what I cannot.

And what I really want to write about today is how that part of the effort is going. I have found, in short, that some of our new tech prostheses are really quite powerful in aiding the only-latently-musical to be fully musical, to manifest our internal music to others. These prostheses are at the same time bringing about a transformation in the way we experience music that is at least as significant as anything that has occurred since the dawn of the recording era. This transformation is exciting, destabilizing, and worrisome all at once. Especially as concerns those tools that rely on artificial intelligence, I have found, technology is giving us something far worse than the “banalities” Brouwer despises.

The music I hear inside of me is, in the end, reprocessed human culture — it is the organic filtering, channeling, and recomposition of the sounds of other human beings, mostly American ones, mostly from the 20th century, sounds that were themselves often, in their initial production, enhanced or vehicled by new technologies, but that continued to testify to a clear origin in the human creative drive. AI music is different. It is, rather, the ground-up and reprocessed waste of mass human culture — the part that got wiped off, and left recordings, footage, tape, film, and data as its trace.

I found my way to Mureka, a subscription-based AI music platform, and I got myself a monthly pro plan. My idea was that I could upload recordings of myself humming or singing these little tunes of mine, so that it might then process them and make them into something appealing. I had expected that there would be a digital keyboard, and that I could supplement or correct any notes that were not entirely to my liking, that I could improve upon the original with small adjustments, some of which might require at least some competent musical input on my part, compelling me, I anticipated, to rise to the challenge and to improve in my musical ability.

In fact I had no opportunity to provide any such thing.

A big part of the problem is that if you upload a tune, the AI can only give you something that is like that tune, since to give you the tune itself, just as you had wanted it, could violate copyright protections on existing songs. Let’s say, simply for the sake of illustration, I upload an audio file of myself singing Miami Sound Machine’s 1985 hit, “Conga”, such as the following: