> Scott developed some of the first devices capable of producing a series of electronic tones automatically in sequence. He later credited himself as being the inventor of the polyphonic sequencer. (It should be noted that his electromechanical devices, some with motors moving photocells past lights, bore little resemblance to the all-electronic sequencers of the late sixties.) He began working on a machine he said composed using artificial intelligence. The Electronium, as Scott called it, with its vast array of knobs, buttons and patch panels is considered the first self-composing synthesizer.
I wish the definition of music were a bit more specific. If you look at it as something that is supposed to communicate an composer-intended emotional narrative, this is still a long way away from that kind of music. I wonder how well AI is doing these days in terms of writing compelling plays, or movie scripts, or short stories? There, the distance between author-intended emotional narrative and work product is a bit more immediate, but in my opinion that is still an essential element even in music.
This is basically sample fragments with some very simple repetition rules, which are then picked and mixed manually.
The repetition makes it sound a lot more musical than some of the more extreme academic music you can find. But it's still a product of its source material enhanced by human editing.
There's no sense in which this is any kind of AI. There's no NN, no symbol grammar, and barely even the simplest rule-based expert system. It's more like a semi-random razor blade tape editor and mixer.
But... it's new to most audiences, who won't have heard of Autechre's more interesting experiments with Max - or any number of other projects - and who are used to a rather precious narrative about the sublime nature of musical composition, which the text shovels on with a back hoe.
It's one of the odd things about music - it's performative in more than one way. Not only are composers supposed to perform music in the obvious sense, they're also supposed to perform "I am a composer and/or performer" in very standardised genre-specific ways.
In academic music and other genres aimed at intellectual audiences that means the music has to be supported - i.e. marketed -
with program notes that suggest to the listener how important, culturally significant, original, refined, and creative the music is.
If you take away the notes and let the music speak for itself, you immediately lose a lot of the impact.
I suspect most people who heard on this Soundcloud without any explanation would think "That was weird and maybe mildly interesting" and move on without giving it a lot more thought.
AI won't have solved music until the music can stand on its own without needing that kind of marketing.
It is better left undefined since that would close its evolutionary scope, excluding great experiments in "noise music". Does it produce something that affects you is what matters.
David Cope's work is interesting in this regard. His algorithms composed sheet music that is played by a pianist. He's done some blind tests where the audience appreciated the "work of Chopin" until they came to know the truth.
"The music was at once futuristic and nostalgic, slightly melancholy, and quite subtle: Even the digital noise samples it used—basically sonic detritus—seemed sensitively integrated."
I had some difficulty taking the rest of the article seriously after reading this. I know music is subjective, but I think the author may have been projecting somewhat.
Did you listen to the samples? They may not be to your taste (Katy Perry aficionados may be disappointed), but I certainly thought they had unexpected emotional tones to them.
Don't forget, this is a three year labor of engineering love for the author!
It is a very subjective thing. I listened to all the samples, trying to find one that I could appreciate. I was unsuccessful.
Before you scoff, I am a trained classical guitarist and play a variety of instruments. To be clear, nobody needs my permission to enjoy it. I found nothing that I enjoyed.
I didn't find anything emotional about them. Other than the concept being interesting, I've not found anything else to appreciate.
That's because the tempo is static, and I don't think there's any "dither" so to speak on the placement of the notes (applying lag and lead, rubato, etc.). I enjoy what the author created a lot, because I like abstract electronica, but it is not so hard to apply algorithmic techniques that create a more lively sound.
That's pretty much the key with this sort of thing. I've done very similar experiments and as long as you choose good source material, even fixed-length samples cut at random and then sequenced rhythmically will sound good.
The author's description of the music is slightly corny but I had a very similar awestruck and eerie feeling after the first time a python script I wrote generated some pretty passable electronic music.
I wonder if people have spent as much time listening to algorithm music as we have reading articles about them. Seems like these approaches generate a lot of press but I wonder how much people actually listen to the music on a recurring basis
Even worse, I can't get the links to open in soundcloud even after installing it... and the app tells you that you can proceed without an account, then soon insists that you need to sign in. So i still haven't heard any samples...
Soon even the pop stars need to fear for their jobs. We just gotta perfect robots to look like young humans make them appealing to their primary demographic and we're good to go.
https://youtu.be/MDIIfAMGLhM
Wikipedia:
> Scott developed some of the first devices capable of producing a series of electronic tones automatically in sequence. He later credited himself as being the inventor of the polyphonic sequencer. (It should be noted that his electromechanical devices, some with motors moving photocells past lights, bore little resemblance to the all-electronic sequencers of the late sixties.) He began working on a machine he said composed using artificial intelligence. The Electronium, as Scott called it, with its vast array of knobs, buttons and patch panels is considered the first self-composing synthesizer.
https://en.wikipedia.org/wiki/Electronium
You will know his popular music from the warnerbrothers cartoons :) https://youtu.be/YfDqR4fqIWE