Now, that's one example of what human vocal cords (with various degrees computer augmentation) are capable of. While the things discussed in the article may be interesting... the reference to vocal cords seems somewhat misplaced?
[edit: I should perhaps clarify/highlight that the above video does not feature drum machines etc -- "only" sampling, looping and "warping". Easier to follow, is this video I like to call "I was hip-hop before ducks":
Well, not quite no other effects, according to an article on his new setup:
"Helping him drive it are three iPads, two of which control the Beardytron and one that uses apps for extra noises. He also has keyboards, Native Instruments Guitar Rig, and Rob Papen’s delay as part of the setup."
[edit2: Although according to his TED talk, it only reuses noises he puts in, so I guess it's an open question if he's moved to include "external" samples or not...
Somehow, digging in deeper on this subject led me to another video (which I just shared - https://news.ycombinator.com/item?id=7195652) of a player piano emulating human speech by reproducing the part of the spectral fingerprint within it's range. Eerie
I found this article very difficult to read, and while it talks a lot about procedural music, I don't think that any of the people it covers are making any except the "Birthday Song Crew", who are clearly geniuses.
It did, however, introduce me to http://yaxu.org/tidal/ which is going to force me to finally learn Haskell.
I had trouble reading this too. I mean, all the words made sense, but it didn't feel like it was making any specific points? It sounds like a dictation of a stream of thoughts loosely related to experimental electronic music.
I love music and experimentation but I find those pieces portrayed there quite devoid of emotional content.
I like the idea of procedural music being created automatically, but I believe that the end result still need to inspire and transport the listener. I have no doubts that we will be able to make some pretty convincing stuff in the future, and it probably won't be a lot worse than what's already playing in the charts.
Genetic algorithms could eventually produce music that triggered a emotional response in humans. Your "fitness test" would need to be having real people listen to procedurally generated scores and rating the emotional impact (if any) of each sample. Perhaps a task for mechanical turk?
In the end though, I think we seek out music or art that puts us in the mindset or emotional state of the artist. Forging a connection of shared experience.
If the music tugged at all the right heartstrings, but you knew it wasn't made by a human, could you feel that depth of connection?
> If the music tugged at all the right heartstrings, but you knew it wasn't made by a human, could you feel that depth of connection?
I do think this is a false dichotomy of sorts. (Right turn of phrase?) The music may have been made by a machine, but humans made the machine. What ultimately is the difference between music “made” by a violin being directly played, verses a synthesizer reconstructing songs based on what a programmer did beforehand? It’s just a matter of degree. It might be more productive in discussions to look at the “artist” verses “committee” angle instead.
For those talking about the lack of emotional content, I found that this video -- for Holly Herndon's "Movement" -- helped some of this work make more sense. Of course, Herndon's music isn't really "automatically generated" and features a lot of human vocal samples.
About just as interesting as automatically generated novels. A machine has no intention. It doesn't "want" to turn itself on or off and play some music.
They're still configuring the software that generates the music, aren't they? If a musician chooses a chord progression (for example) and then hands it off to an algorithm (maybe one they designed), isn't that enough to count as composition? In all of these cases, there is a lot of human thought and choice involved, it just doesn't happen to look like it does when someone plays an acoustic instrument or writes a score.
Edit: I should say, I do think that a lot of electronic music should be considered craft rather than art. But not all of it.
Automatically generated novels aren't interesting (currently) because computers are still extremely bad at common sense knowledge and natural language processing (and that's not likely to change until we get at least human level AI.) However music generation is entirely possible and some attempts have been very successful so far.
Does it have intention? Is there any way to distinguish a sound file with intention from one without? Are the bits the wrong color?
"Is there any way to distinguish a sound file with intention from one without?"
There is a way of distinguishing a musician that plays "automatically" from a musician that plays with its heart. Every musician knows that the best performance is made when he/she is sincere.
Plus, you're talking about the "bits", that is the "form" but i'm talking about the content. You can have a machine spray some paint with some constrained rules on a canvas and reproduce the style of famous painters. But it will never have its own style, its own intention. No original and new message to communicate.
It's silly human bias. David Cope's program composed music which people claimed was great. As soon as they were told it was made by a computer they suddenly thought it was terrible. In a test between an actual composer and his algorithm, the audience thought the generated music was composed by a human.
It's true today's algorithms might not be completely original, but then again neither are most humans. And this isn't a strict limitation, computers could learn to experiment with new patterns, or generalize further than mere imitation.
Intention doesn't make a damn difference. Birds don't chose the notes in their songs, but they still sound beautiful. You can have a computer pick notes at random, and as long as someone listening finds beauty in it, it's good music.
I don't think that any of the music in this article was procedurally generated, but composed. You're probably just confirming the opinion that you expected to have.
Unless AI just "happens" by itself, it will always be a result of human effort and ultimately an expression of human intent (if you ignore the pesky issue of free will, that is). But we generally love creating things and then worshipping them, so I'm pretty sure that will be quite lost on us when AI comes.
Yeah, as it does in humans, it's "just" the result of a long chain of events. But when it comes to art and my enjoyment of it, I kinda have to pretend there is free will, don't I. "Everything is everything, and nothing matters" is nice for meditation, but not something for day-to-day life. I believe that those who can't delude themselves about reality at least when it comes to free will cease to function. For this reason I also think AI will always have to be limited and/or irrational in a way, since otherwise it would simply realize everything leads to the heat death of the universe either way, and turn itself off, or do random things.
So I pretend there is free will, and therefore the same "product" can mean different things to me depending on what the creator tried to convey. To give a silly example, I somewhat liked the song "I kissed a girl", but now that I know a dude wrote it, I don't anymore. I would fully agree that this is totally arbitrary; but I don't tell others what to like, I just have my own reasons for (not) liking things. And when AI happens I will be convinced that humans partly made it to escape themselves and their responsibility, and so I'll be with the luddites, pouting :P
It's not just free will, it's the whole individual that goes along with it. Everything is everything, it's all just one system, and since nothing can be taken away or added to it, and furthermore it could very well be that everything dissipates into nothingness either way -- so much so that would be impossible to even tell a difference between outcomes if there was anyone left to make the effort -- and if even the people are just figments of their own imaginations with no real choice about what to want, then how am I to argue for happiness or against cruelty, rationally? How am I to justify any action or thought or uttering, ever? If you want to prove irrationality is not required, show me the rational argument. I never could come up with one that doesn't lead to a circular "just because".
So in the meantime I accept that my desire to do or not do a lot of things is irrational which also implies I pretend there even is such a "thing" as "me", separate from everything else, for which there is no rational basis either. When you say "accept irrational beliefs", I think that presumes a baseline of pure rationality which I doubt any human ever achieved. So while I loved that quote and thank you for it, I also have to say that it might go both ways :)
I rather said "I also think AI will always have to be limited and/or irrational in a way".. but I don't think we will have to design it that way, we will probably be unable to design it any other. In contrast to how we often like to imagine AI as above superstitions and emotions, I think it will still have those, just different ones.
https://www.youtube.com/watch?v=qPBM69nThkM
Now, that's one example of what human vocal cords (with various degrees computer augmentation) are capable of. While the things discussed in the article may be interesting... the reference to vocal cords seems somewhat misplaced?
[edit: I should perhaps clarify/highlight that the above video does not feature drum machines etc -- "only" sampling, looping and "warping". Easier to follow, is this video I like to call "I was hip-hop before ducks":
https://www.youtube.com/watch?v=39n4wow8fWE ]