Sound-Alikes: Litigating the AI Mimic
Dani Deahl in The Verge:
The word “human” does not appear at all in US copyright law, and there’s not much existing litigation around the word’s absence. This has created a giant gray area and left AI’s place in copyright unclear. It also means the law doesn’t account for AI’s unique abilities, like its potential to work endlessly and mimic the sound of a specific artist. Depending on how legal decisions shake out, AI systems could become a valuable tool to assist creativity, a nuisance ripping off hard-working human musicians, or both. […]
If [an AI] system then makes music that sounds like Beyoncé, is Beyoncé owed anything? Several legal experts believe the answer is “no.” […] “There’s nothing legally requiring you to give her any profits from it unless you’re directly sampling,” [Public Knowledge policy counsel Meredith] Rose says. There’s room for debate, she says, over whether this is good for musicians. “I think courts and our general instinct would say, ‘Well, if an algorithm is only fed Beyoncé songs and the output is a piece of music, it’s a robot. It clearly couldn’t have added anything to this, and there’s nothing original there.’”
I’m not so sure. It could turn out that the controversial “Blurred Lines” ruling laid the groundwork for litigating AI-mimicry.
Micro-licensing sites — where you add your music to an online library portal and licensing rights are granted for a small fee — have an allure. For one thing, they’re convenient as you merely upload your songs and fill in some info. The only other step is to cross your fingers and hope your music takes off on the site. Micro-licensing sites promote royalty in scale — that if your music is successful, it won’t matter that the license fee is $15 (or whatever). Thousands of those will add up.
But it’s common knowledge that it’s a 1% of 1% that have such success on these sites. And I’d wager these successful composers work super-hard at it, doing quite a bit more than crossing fingers.
You also have to accept that, in most cases, you’re losing complete control of your work when you supply music to one of these content providers. You don’t know who’s licensing, and you certainly can’t deny a license, and you won’t be able to gauge or benefit from a commercial entity making loads of money off your work.
Stock photo sites operate the same. Which brings us to this story reported in Petapixel:
It turns out a Newfoundland-based company called Islandwide Distributors (IWD) had licensed [Michael] Stemm’s photo royalty-free from Shutterstock for just $1.88. The company then turned around and made at least 500,000 units of products with it [and sold the products in Walmart stores across Canada] — Stemm learned this number after reaching out to the company. So while Stemm’s experience may seem unfair, it was likely entirely lawful and within Islandwide’s rights.
The salt on the wound:
Unfortunately for Stemm, he isn’t even able to withdraw the $1.88 he earned, as his account needs to reach a balance of $50 before he can see the funds.
Techdirt has no sympathy for Stemm:
Stemm said Shutterstock could license the photo. Shutterstock did exactly that. The fact that Walmart has more than 500,000 items featuring Stemm’s photo is probably unexpected, but if you really want to retain full rights to your creation, you don’t hand part of those rights over to a middleman. When Walmart licensed the picture from Shutterstock, it didn’t seek Stemm’s permission because it didn’t need Stemm’s permission. […]
It certainly seems unfair when a company can make hundreds of dollars from a $1.88 license. But there’s nothing unfair about a process that involves a voluntary relinquishment of control. Shutterstock can certainly find a greater market for someone’s photos, but no one should go into this relationship believing it will result in newfound personal wealth.
I agree with Techdirt’s sentiment here — when you enter into an arrangement with a stock photo library, or a music library, or the record label that will own the rights to your songs, you need to accept what you’re getting into. Moving forward, any mistake is but a learning moment as you got yourself into the mess. According to Petapixel, even Stemm admits he didn’t read Shutterstock’s licensing terms before clicking ‘submit.’
That said, I do believe these sites could do a better job explaining what’s in store for content creators and to favor realistic expectations. The idea that an artist will make ‘easy money’ through a micro-licensing site supports a rare exception. The artist might have better luck buying lottery tickets. Just like the lottery, the allure is strong — I’ve certainly been tempted, by both these licensing libraries and the lottery. And it’s okay if you play — just understand the agreement that you’re entering, the potential outcomes, and how frustrating it is to be in Michael Stemm’s position.
There’s a misconception that sampling has a time limit, that one can legally sample anything (a drum hit, a vocal yelp, a guitar riff) as long as it’s short. I’ve heard various guidelines set for this assumption with the two-second mark as the most common. It’s not true. If a judge or jury can identify a sound — any sound — as originating from a copyrighted source, you’re probably in trouble.
Granted, the US courts have not been consistent in how they rule on this, and it’s fair to say there’s no set precedent here. But, in the EU, this may no longer be the case thanks to repeatedly snubbed Rock-and-Roll-Hall-of-Famers Kraftwerk and the synthesized sound of banging metal.
The case involves a two-second sample from “Metall
aufMetall,” which the producers Moses Pelham and Martin Haas used as a continuous background loop in the 1997 song “Nur Mir.” Although the sample consists of just two seconds of the original song, it’s recognizable and important in “Nur Mir,” which is performed by the singer Sabrina Setlur. […]
… Advocate General Maciej Szpunar advised the European Court of Justice, which is deciding a copyright case that involves Kraftwerk’s “Metall of Metall,” that even limited sampling of a recording can constitute copyright infringement. Advocate General opinions are not binding, but they’re watched
closely,since they often predict the way the high court of Europe will decide cases.
The case is interesting as it involves only the recording (master) side of the music and not the composition (written song). The idea is that one can take a ‘sample’ of a written song in isolation — such as a
Basically, when you sample a two second clip of a track, you are sampling both the recording and the song contained within it. But it might be hard to argue that the two second snippet of the song can be protected by copyright in isolation. However, at the same time you could argue that the two second snippet of the recording is.
The ‘Metal On Metal’ case centres on the recording rights. In essence, in the 2012 court hearing, when one argument on the Kraftwerk side was that Pelham could have recreated the sounds he sampled, they were basically saying that there was no song copyright to infringe here, but that the separate recording copyright had been infringed by the uncleared sample.
In his ruling, the Advocate General wrote, “A phonogram is not an intellectual creation consisting of a composition of elements such as words, sounds,
Some of the questions referred to the European Court of Justice involve details of European law, including whether the German concept of free use is compatible with EU law. But others get to the center of the debate around copyright and free expression. Generally, most countries’ courts have held that quotation doesn’t infringe copyright when a new work refers to the original one – in a book review, or even in a parody of a song. But what about when the new work has nothing to do with the original? Such questions have increased urgency in the digital age, and this is one of several important cases on the topic.
I wonder if this will create more consistency in how the courts rule on ‘short sample’ cases in the US. And I also wonder if New Order ever had to pony up for their own short Kraftwerk sample in “Blue Monday.”
It’s finally over, but this probably won’t be the last post about it. Via Forbes:
The five-year copyright battle surrounding the popular single “Blurred Lines” finally ended this week with a judgment of nearly $5 million against artists Robin Thicke and Pharrell Williams. Marvin Gaye’s family had accused the duo of infringing Gaye’s 1977 song “Got to Give It Up.” California Judge John A. Kronstadt’s final ruling, which entitles the Gaye family to half of all royalties to the song moving forward, effectively puts an end to the highly publicized lawsuit. […]
Thicke and Pharrell … failed to petition the U.S. Supreme Court for a writ of certiorari by the applicable deadline. As a result, the judgment has been entered against them, and the case is now effectively closed.
The controversial “Blurred Lines” copyright infringement case is officially precedent. If there’s any silver lining for Thicke & co., it’s that the Marvin Gaye estate only gets 50% of future royalties. I’m surprised they didn’t get 100% (which is not out-of-bounds in infringement rulings).
In the end, the most interesting court case related to this ruling won’t be the “Blurred Lines” affair but what ends up being the first similar high-profile copyright lawsuit to follow. Who will it be? Led Zeppelin vs. Greta Van Fleet? The Fall vs. Pavement? The Bob Marley estate vs. every contemporary reggae artist?
I kid. But the test will be how far someone wants to take this precedent. In an age when there’s a lawsuit over the lyric “haters gonna hate,” you can bet the limits of the ruling will get tested sooner than later. I’m not even sure how a jury could reach a sensible conclusion, as similarity in style is arguably subjective. I have a mental image of a bunch of stoners sitting around a turntable, and one says, “man, this sounds just like The Piper at the Gates of Dawn,” and everyone agrees.
There’s a fine explainer in The Vulture with copyright expert Jeff Peretz:
… the issue at stake in this case is the “how,” not the “what” — in other words, how the song is presented. Pharrell admittedly set out to create a “party” song in the style of “GTGIU” knowing full well it was within the legal boundaries to do so. He, in essence, walked right up to the legal line but made sure not to cross it. What the original decision did was move the line. By letting it stand in appeal, Judge Milan D. Smith Jr., who wrote the majority opinion, has literally changed the rules. But if the difference between “what” and “how” were properly explained to the jury the first time around, this never would have gotten this far.
Law professor Edward Lee argues in Billboard that the “Blurred Lines” sonic homage to Marvin Gaye could have fallen under ‘fair use:’
I’m not sure which account best explains what happened. But the jury believed Pharrell and Thicke copied from “Got to Give It Up,” and hence the verdict. What would have happened if Pharrell and Thicke invoked fair use and told the jury that they borrowed a small part of the ’70s groove from Gaye’s song, but gave it new meaning, a different character and new expression in a modern pop, dance song fitting today’s tastes? Surely, no one would mistake “Blurred Lines” for a song from the ’70s — it has a much different character. Of course, the jury might have rejected that defense as well, but fair use fits better with Thicke’s original explanation in GQ that the duo intended to borrow the groove of “Got to Give It Up” in creating their new song. Moreover, having the jury or court balance the factors of fair use provides a more nuanced and realistic assessment of the competing interests at stake than the simple test of infringement does.
He adds: “The music world needs a clear decision applying fair use to a non-parody musical work. Without it, copyright lawsuits will only increase.” It’s no surprise that I agree. The democratization of distribution, in turn making almost every artist work available to the public, is leading to an overflow and breakdown of copyright litigation. No artist creates in a vacuum, and if we want to continue to creatively ‘Steal LIke An Artist,’ then something’s got to give.
America’s Copyright Royalty Board yesterday got around to thinking about what the country’s mechanical royalty rates should be for the next five years.
Mechanical royalties – paid to songwriters when recordings of their songs are copied and distributed – are covered by a compulsory license Stateside. Which means songwriters and music publishers are obliged to license third parties making and distributing those copies at a statutory rate, so that rate-setting processes like this one are rather important.
Traditionally the main customers of mechanical rights have been record companies, which need a license from the relevant songwriter or music publisher every time they press a CD.
In the US, unlike in Europe, it was the label which paid the mechanical royalties on downloads too, so that iTunes didn’t have to worry about making sure the owner of the song copyright was paid their share of any income.
However with streams, where both the mechanical and performing rights of the copyright are exploited, it is the digital platform that is the licensee and which therefore pays the mechanical royalties directly to the writer or publisher (or not as the case may be, as those songwriter lawsuits against various streaming services have demonstrated).
Discs and downloads also remain a decent part of the recorded music business for now of course, but – after a bit of a stand off – the US record industry reached a deal with the music publishers on mechanical royalty rates last year. Which means that the CRB hearing is very much focused on the rates paid by the streaming services, which are, after all, where all the growth is in recorded music these days.
The tech giants are expected to argue to reduce the amount they pay, while the National Music Publisher’s Association and the Nashville Songwriters Association International will lobby for an increase.
NMPA wants songwriters to be paid each time their song is played, or each time a user purchases a subscription. It also wants to share the profits from the sale of technology and subscriptions that include access to music.
The US government has been setting mechanical royalty rates for over 100 years, beginning in 1909 when Congress determined that the rights would be subject to a compulsory license. This means that anyone can record a songwriter’s work for a fixed rate without permission or approval. Congress used to set this rate, but has since delegated the task to the CRB judges. The current rates were set over ten years ago when digital streaming was just starting to take off.
With the release of iOS 10, song lyrics are now displayed within Apple Music. Apple have received incredibly positive feedback from members, who can now follow along during playback of their favourite songs. To ensure songwriters are paid Apple is obtaining the licenses required to display lyrics in Apple Music. Apple rely on accurate songwriter and composer data to efficiently obtain these licenses.
Apple says to “make sure the ownership of your song is registered with a publisher, and that they have registered ownership with relevant publishing agencies such as ASCAP, BMI, PRS, Harry Fox and Music Reports.” That obviously is misleading.
First of all, we can’t be that surprised that Apple has this impression because as we all know, it is frequently lost on HFA and MRI that neither of them is in fact the government. However, given that Amazon, Google, Pandora and others are sending millions upon millions of NOIs to the Copyright Office claiming to have no idea who owns songs by very well known artists, it should make it obvious that the one place you need to “register” your song copyright ownership is with the U.S. Copyright Office.
It’s also misleading to state that you have to have “the ownership of your songs…register[ed] with a publisher” which may happen frequently, but is not required to enjoy ownership rights.