I became the world's most prolific DJ, using code
royvanrijn.com
royvanrijn.com
I think Adam Neely makes a more relevant point in his videos about Dark Horse and Levitating: it doesn't really matter what any actual infringement claim is because you can typically find relevant prior art from legitimate works of music if you dig deep enough, even without going through the exercise of autogenerating note sequences.
As I understand, the legal arguments focus on whether there is a clear and traceable connection between the creative process for a song and the alleged infringed work, and whether there is clear intent to omit credit where it is due. I.e. the argument already starts from the assumption that similarities and inspirations from existing works of art can and do exist.
There's a bazillion examples of flat out copying being considered perfectly fine: the lyrics for Anne Marie's "2002" clearly uses iconic lyrics from Britney Spears pretty much verbatim, and in a way that is obviously intended to refer to Britney, but the thing is that many many other artists before her have done similar homages (e.g. Calvin Harris' "My Way" references Sinatra's, etc) so there's a fairly strong reason to believe that a lawsuit based solely on similarity grounds would go nowhere.
The reason this distinction matters is because I think its important to understand that not only European musical traditions will have equivalent rhythmic complexity but also other musical traditions will have equivalent harmonic complexity. Just because we don't have a great model of other musical traditions' harmony, does not mean they lack harmony. And vice versa. E.g. during Baroque era although temporal information was rarely denoted on paper (any more than 3 time, 4 time, tempo etc) musicians performing these pieces had to express a certain understanding of rhythm. Pieces were never played like MIDI, they always had rhythmic nuances.
Karajan might have been the last of the conductors that I enjoy like listening to. His fifth changes pace but it feels so natural. The slow parts are drawn out were they need to be, but the fast parts just grab you and drag you along. There was no notation for that, it took interpretation. And he could get all the instruments to open the piece together. Even Berenbaum couldn't get his players to open the piece properly, at least not those that I could find on Youtube.
Thought there is one young guy on Youtube who does a terrific job conducting, I should go find that.
An AI generated song machine would have to nail a lot more elements than just the melody notes to properly stop music copyright cases. In my view, a more interesting project that might be more effective in defusing lawsuits would be to try to catalog all of the musical tropes that define genres, then attempting to detect how common they are in that genre. In an ideal world, maybe this would be able to drive a metric of how similar specific two songs are vs. picking any two songs in that genre at random.
He most certainly did not. Of all the different takes out there, his is very weak.
>Most recent music infringement lawsuits seem to argue that some combination of...
There is a very good reason: as he mentions, the chords diversity use in pop songwriting is typically so poor that based only on that, the amount of things considered plagiarism would thus be ridiculous. If the similarities affect almost all dimensions (style, arrangement, rhythm, melody, ...) to the point of being "essentially the same", then it's exactly what people would want the law to exist for.
I think Adam Neely did a good job explaining what infringement lawsuits mean in the context of popular music production. Whether or not you agree with the strength of his case on this particular lawsuit, well that's not quite the point I was trying to make here. Still, what do you consider to be a strong take on this case?
And in general, you must be "creative" to have a copyright. You might have a "copyright" on the resulting file, but no court would ever dream of extending that to a claim of copyright on every melody. There is no way that the author was "creative" in any sort of proportion to the amount of material being putatively claimed. Normally one would expect for this to then start a big HN chain arguing the precise definition of "creativity per unit output" that is the threshold, but in addition to the fact you have to get a court to agree to your definition, bear in mind that this is quite literally exponentially little creative effort per output. The author has put in so little effort per output that they haven't even listened to their own "work" once, I'm sure! This is not the normal definition of a "creative work". The usual arguments will be based around a polynomial at most, and frankly usually linear amount of output per "creative input". Especially in light of the fact that the lifespan of a given "creative human being" isn't even "linear" so much as "constant".
Given that this is only a very small amount of effort from being able to claim all combinations of notes ever, it's clear this is not a copyrightable work, excepting perhaps the literal output of the work but no more than that.
(It's also not that much more work to order these things in entropy order, by analyzing songs and deriving some probability for note lengths and intervals, making it so that one could just start generating melodies and actually hit almost every useful melody in an even "smaller" work. Also not copyrightable.)
So, basically, don't learn anything about copyright from this article.
I was doing legal research on this, and funny enough there is an article that discusses this very issue: http://mttlr.org/2016/11/no-mans-skynet-copyright-in-procedu...
or if i create a completely different procedural generation method for planets and 99% of my planets differ from no mans sky but there is an identical overlap of some of my fringe planets and theirs, it again does not feel like copyright.
but if i take their method and tweak it a little bit so i also make a game with 18 quintillion planets of a somewhat different variety but am only able to do so because i am copying their methods without their consent.. that feels much more like copyright. even if none of my planets are an identical match to a single one of theirs
You don't generally end up with a copyright on all possible combinations. You may still de facto "own" the copyright if all possible combinations encompass parts of your own creation. For instance, I've been playing XCom 2 lately. It has a character creator. Firaxis can't claim to own all the possible combinations of characters it can create, but they own a substantial portion of the parts. I don't think they could claim they own eyes of a particular color or particular shades of skin, but all the clothing, accessories, tattoos, guns, etc. all individually have copyright, so it's still not like I personally can just crank up the character creator and claim a copyright on some particular one and start using it for whatever commercial purposes I desire.
No Man's Sky, from the looks of it, is in the same boat. They don't necessarily get "a copyright" on everything their algorithms can possibly generate, but at the same time, if someone produces an exact match of any of the 18 quintillion planets they must be using plenty of copyrighted assets along the way. So in practice there may not be a big difference.
Where the difference comes in is when the pieces get to be so fine that they are not themselves copyrightable. To put it in a visual context, the original favicon format was 16x16. Even at full 24-bit color, that's only 2^32 possibly favicons. It's trivial to enumerate them. But you can't copyright a single pixel, and you can't simply claim a copyright on all combinations. The former is a bare fact, and the latter had no creativity ("enumerate all possibilities" is not creative, it's a homework assignment in Comp Sci 201). You can't copyright a single note, it's a bare fact. You can't copyright all enumerated combinations of them. You can copyright the program used to generate them, but that doesn't give you rights to the output.
So, they do and they don't. And the sense in which they do doesn't match the sense in which trying to copyright all possible melodies does.
In the event that you have something like https://www.youtube.com/watch?v=8fZBUsn5RYg , kkrieger, a super-procedurally-generated game that doesn't have any clear textures or geometry in its code, I think you could claim a copyright on what gets manifested, but not on every conceivable thing the algorithm could generate. What gets manifested will be a much more reasonable amount of protection relative to the effort, merely polynomial at most, rather than exponential. Exponential is, you know, really big.
However, people will sometimes claim that a portion of their work is stolen.
The idea here is to give someone a citation of it existing elsewhere as a defense to the above. Although this is a very experimental endeavor and some suspect it won’t be taken very seriously in an actual court, because the context in which it was created may undermine some of the creative requirements for qualification under copyright law.
I understand why they want to take back the copyright on music, but they do so in such a geeky way that it seems completely useless to me.
Ultimately, musicians will pick good / cool melodies from this dataset, in the same way they do when in front of an instrument.
I might be missing the point ?
As a musician, I think what would happen is the companies that abuse copyright now would keep abusing musicians, except now they can just take any song they like without compensating the person who made it.
The melody copyright is where things get really hazy. It's hard to determine when a melody infringes on another melody. What if the notes are the same, but the timing is different? What if the notes aren't exactly the same, but are pretty similar? What if the main melodic ideas are really common in a given genre or style? How do you determine if there was actual copying, or if two musicians just came up with the same idea independently? What if the melody just isn't an important part of the style of a given work? What if the melody is almost the same, but used in a completely different musical context?
There isn't an objective set of criteria that can determine if a melody infringes on another melody, without being too narrow or too broad. And since there's no good criteria, the only way to litigate this is to have better lawyers then the other guy, which rarely works out well for independent artists.
[1] At least, uncontroversial at a high level. The details get messy really quickly.
https://www.buzzfeed.com/reggieugwu/what-the-law-says-about-...
> Copyrighted elements of a musical composition can include melody, chord progression, rhythm, and lyrics — anything that reflects a "minimal spark" of creativity and originality.
https://ask.audio/articles/5-common-beliefs-about-song-copyr...
> Song Titles and Chord Progressions are not copyright protected
> This is true.
[snip]
> Ditto for chord progressions. There must be hundreds of songs that were hits in the ’50s and early ’60s that followed the familiar “ice cream changes” progression of I-vi-IV-V7. Thin “In The Still Of The Night” (another song that shares its title with others), “Donna”, “Silhouettes”, “This Boy” just for starters. Also the chords from the “Pachelbel Canon in D” have been used numerous times. Think “A Whiter Shade Of Pale.”
Does lack of enforcement make something legal?
https://moviecultists.com/can-chord-progressions-be-copyrigh...
> "If a single chord progression were elaborate enough and unconventional enough, it could be protected."
How many possible chord progressions are there?
> We had initially considered a “de Bruijn” sequence. But if we were to use a single file, that would have down sides:
> If someone infringes our work, it would only be a tiny percentage (0.0000000001%?) of the “work” — so someone would argue “fair use”
> Same idea with others incorporating ATM works in theirs (“tiny percentage”)
> So our technical/legal design is “One MIDI file per melody” — which I think is a legal feature, not a bug.
"Thaler listed Creativity Machine as the author of the work"
You cannot claim that the AI was the creator.
Which makes a lot of sense to me - Someone had to set up the automation with the intent to create a copy writable work, and copywrite often has built in expirations based on a window of time after the death of the author. Hard to make that sane if you're listing a computer program as the author - when does it die?
https://en.wikipedia.org/wiki/Feist_Publications%2C_Inc.%2C_....
People keep throwing that article around, and there seems to be a profound misunderstanding about what was determined there - automation is fine. Listing a machine as the author is not.
In contrast, a musical melody is created by a finite set, and is bracketed by a duration. You can't keep adding more notes to a composition without extending its duration. Doing so makes it non melodic. If you do it enough, it becomes noise (white noise, pink noise..)
But imagine I build something that spits out as many binary sequences as possible. Do I then have a copyright to all the "works" that can be interpreted from it in various data formats I may have accidentally met?
This question is about intention, not authorship.
The court is saying “the [typewriter] didn’t write the book, you wrote the book using a [typewriter]”
Replace [typewriter] with any technology.
http://www.spiderrobinson.com/melancholyelephants.html
--- start quote ---
There are eighty-eight notes. One hundred and seventy-six, if your ear is good enough to pick out quarter tones. Add in rests and so forth, different time signatures. Pick a figure for maximum number of notes a melody can contain. I do not know the figure for the maximum possible number of melodies--too many variables--but I am sure it is quite high.
"I am certain that is not infinity.
"For one thing, a great many of those possible arrays of eighty-eight notes will not be perceived as music, as melody, by the human ear. Perhaps more than half. They will not be hummable, whistleable, listenable--some will be actively unpleasant to hear. Another large fraction will be so similar to each other as to be effectively identical: if you change three notes of the Moonlight Sonata, you have not created something new.
I do not know the figure for the maximum number of discretely appreciable melodies, and again I'm certain it is quite high, and again I am certain that it is not infinity
--- end quote ---
Do read it It's scary and prescient.
https://news.ycombinator.com/item?id=22301091
Ex. Once a work of song makes $10M it should just go into the public domain. Want to make more? Keep writing more songs.
https://www.rollingstone.com/feature/in-the-jungle-inside-th...
Tom Johnson's Chord Catalog which organizes the 8,178 chords possible in a single octave of the piano: https://fr.wikipedia.org/wiki/The_Chord_Catalogue
James Whitehead's All Possible CDs: http://www.jliat.com/APCDS/index.html
> This “thought” experiment although based on real “physical” objects can be treated as a simple mathematical object and so allows us to explore some of the consequences of this object or objects. The important feature is that any finite series is fixed, so greater sized disks, blue ray, whatever, is not significant to the idea, that is in a finite universe there are a finite number of finite objects. The size of the bit strings set real limits on the number of possible objects; web pages typically use 24 bits to encode colors, 8 bits for red, 8 for blue, and 8 for green that gives 256 x 256 x 256 or 16,777,216 possible colors, and no more.
> In Deleuzean terms, you could call this, all possible CDs, the “virtual plane”, thought experiment, in the case of 2 to the power 6265728000 of all possible audio on CD, a virtual set of possibilities or a virtual plane, and the actual physical CDs in the world are actualizations of these virtualalities. Actual objects, physical CDs, being intensities on this virtual plane. Actual CDs are not mere copies of there virtual counterparts, they are not re presentations of the virtual, for they have many more properties, many physical properties, color, size, shape etc., just as in the Deleuzean Virtual and Real planes, the real is not a copy of the virtual, but an intensity.
> Using this as a model we can “experience” actualities that are physically unlikely for humans if not in practice impossible, for 2 to the power 6265728000, is approximately 10 to the power 2000000000. There are only 10 to the power 118 particles in the universe so a full and total actualization of the virtuality of CDs seems impossible.
All possible scales with 12TET and octave equivalence.
The whole story reminded me of this axis of awesome (OZ comedy group) song (well worth watching): https://www.youtube.com/watch?v=5pidokakU4I&list=RD5pidokakU...
With code you really can do "a thousand monkeys at typewriters", but is it meaningful?
> They generated all songs of length n=12 with k=8 notes, this amounts to a staggering n^k = 8^12 = 68,719,476,736 unique songs.
> All these songs are 12 notes long and have their own MIDI file which adds even more overhead. The size of this dataset is huge, 1.2TB compressed using GZIP.
Now, you're correct if you're saying that not all music has melodies. Or that things like Indian ragas would be difficult to run through exhaustive permutation. But, as far as I know all music in the world has native notation or can be notated and reproduced.
I’ve never heard of Him, and I run in those circles.
Interesting reading though.
Edit:
pro·lif·ic (adj)
marked by abundant inventiveness or productivity
Also you really just made a throwaway to insult someone's article?
While I respect the author for the goal here, I would say he was neither abundantly inventive nor productive.
His inventiveness boiled down to taking an existing concept, and saving some disk space using an already known algorithm. To boot - his method of saving that space rendered the original intent (preventing copywrite of melodies) useless, and was already considered by the original group.
His productivity was a single morning's work, creating a single album, which is just random noise and will never be meaningfully listened to even a single time.
I'm not going to begrudge him the usage in the title, and I think it's a fun weekend project, but I also think it's pretty clearly clickbait (which is fine!).
"Productivity is the efficiency of production of goods or services expressed by some measure."
An whole album with millions of melodies in just one morning is quite efficient :). Meaningful or deep, maybe not. But productive... yes!
"There are many different definitions of productivity (including those that are not defined as ratios of output to input) and the choice among them depends on the purpose of the productivity measurement and/or data availability."
So really what we're arguing about is the "purpose of the productivity measurement". I don't think "number of melodies that will never be listened to" is a particularly good measure for "DJ", but it is a great measure for a blog title.
No knock on his accomplishments
Its amusing, it got me to click though and I was pleased by what I read