GrindEFX

GrindEFX is back, w00t! Over at Techdirt, Michael Masnick highlights an interesting development in machine-generated music: a company claims to have created software that can recreate musical performances by famous musicians, living or dead.

As described on the blog, “Zenph Studio’s approach is to work out how the musician and the instrument acts and responds, then get a computer to play that track again as a real-time, real-life performance, which in turn can be recorded using modern techniques.”

It’s essentially MIDI on steroids. Zenph claims that through its software process, it can capture the nuances of any performer’s style and translate that into an algorithm to generate new recordings of old songs – or make entirely new songs that sound like they were performed by a famous musician.

Leaving aside whether this is a good idea in the first place – how long before we have Kurt Cobain hawking deodorant? - Masnick concludes with what I can only describe as an excellent law school final exam question -

So, now, take this software that supposedly can perfectly mimic a certain musician’s playing, and have it record a song. Say it’s a new song. Who owns the copyright? What if it’s adding John Lennon to a Rolling Stone’s song? Who owns the copyright? What if it’s an old song, updated in some slight way? Who owns the copyright? What if it’s just the same song but “remastered”? Who owns the copyright? The legal questions raised by this kind of software are going to keep copyright lawyers busy for a long, long time.

Being the slightly sadomasochistic law student I am, I couldn’t resist analyzing the issues raised in this situation. I am not a lawyer, and this article is only for general informational purposes; nothing here constitutes legal advice.

Who owns the copyright in a new song created by this software?

So, there’s this girl you like, named Ann Tates. You, being a musician, want to write a song for her (”She makes me pancakes!”) Her favorite artists are Jerry Lee Lewis, Jimi Hendrix, and Yo Yo Ma. Let’s say (assuming the software eventually does this type of thing) that you create your new song with this software.

The song itself is entirely new, including all music, lyrics, and backing parts. But the piano part sounds like it’s being banged out by Lewis, the guitar sounds like something Jimi would play, and the cello part has the characteristics of a Yo Yo Ma performance.

As far as ownership of copyright is concerned, nothing in this situation changes the fact that you, as the musician, wrote and composed the song, so you would own the copyright. But let’s make it more interesting and say the computer program is more like Jammer or Microsoft Songsmith. Instead of actually writing the song yourself, all you have to do is pick the performers you want to emulate, select a music style, and hit ‘go’ – the program does the rest of the work.

Who owns the copyright now? Copyright protection does require only the barest “modicum of creativity“, but here all you’re doing is pressing a couple buttons – theoretically you could do this all day and generate millions of songs.

Intellectual Property attorney William T. Ralston explores the answer to this question in his 2005 article “Copyright in Computer-Composed Music: Hal Meets Handel.” (52 Journal of the Copyright Society of the USA 281). Noting the struggle that copyright law has had with new technologies since the very first Copyright Act, he points out that “no direct legal challenge to the status of copyright in a machine-generated musical work has been raised” to date. He comes to the conclusion that yes, a computer-composed musical work would be protected by copyright, and the user of the program (not the programmer) would typically be the owner of that copyright.

What if it’s adding John Lennon to a Rolling Stone’s song?

Remember that when you’re talking about recorded music, you’re dealing with two entirely separate copyrights – a copyright on the underlying music composition, and a copyright on the sound recording itself. While both copyrights might be held by the same author (for example, an independent artist who records her own songs), it’s important to note that copyright protection in one of the copyrights doesn’t extend to the other.

That is, you’re free to record yourself playing a song in the public domain – Beethoven’s 5th perhaps – and you would receive copyright protection on the sound recording, but that doesn’t stop anyone from recording their own version of Beethoven’s 5th. Your copyright protection on the recording doesn’t give you any exclusive rights to the underlying composition. The same is true for cover versions of songs – anyone and everyone can make their own recordings of a track. None of them have any claim to the underlying musical composition though. And because of the scope of copyright protection on sound recordings, none of them can sue someone else who made a recording of the same song for copyright infringement, no matter how similar the two recordings turned out.

Just like Sir Mix-a-lot, copyright law likes big buts. The ‘big but’ here is that while you get a copyright on a sound recording of someone else’s musical composition, along with the exclusive rights copyright gives you, you still need permission from the composition’s owner to reproduce, distribute, publicly perform, etc, (I’ve truncated the ‘etc’ in this list for simplicity purposes) your sound recording, since you are simultaneously reproducing, distributing, or publicly performing the songwriter’s copyrighted work.

Getting back to the original question, the answer is clear. Since the software is generating its own sound recording (rather than ’sampling’ the sounds from any existing sound recording), you would own the copyright on a sound recording of John Lennon playing on a Rolling Stone’s song. You would not, however, be able to do anything with that recording without the permission of John Lennon or the Rolling Stones (or whoever happens to own the copyright on the musical composition you’ve used).

What if it’s an old song, updated in some slight way? What if it’s just the same song but “remastered”?

The same answer for the previous question holds true here. Again, so long as you’re not actually using the sounds from the original sound recording itself, you would own the copyright on a new sound recording you create from an old song, but the owner of the copyright to the musical composition still retains his or her rights to the song itself. You would need permission from them before distributing, reproducing, or publicly performing your sound recording (or not, if you qualify for a statutory compulsory license).

If you have made changes to the original composition’s  ”basic melody or fundamental character,” you may need additional permissions from the composition owner for creating a derivative work. (’Remastering’ or altering the arrangement to the  ’extent necessary to conform it to the style or manner of interpretation of the performance involved’ does not create a derivative work.)

You may be fine under copyright law, but there are other laws too

Masnick was only asking about the copyright issues that come into play with the software described here. Appropriating an artist’s “style” of playing may implicate various state and federal laws concerning right of publicity or unfair competition, to name a few.

Bette Midler, Nancy Sinatra, and Bert Lahr are just a few of the celebrities who have sued after discovering someone else had used a “sound-alike” to create a new work that mimicked their vocal styles. Copyright doesn’t work here – you can’t copyright a vocal style. Instead, they sought remedies through other laws, including those mentioned above.

These types of claims are complex to analyze. Tort law varies from state to state – what might be a valid claim in California may not be in Wisconsin – and the elements required to plead a case under each law differ. And while you can’t copyright a vocal style, you may still find yourself losing your case if the claim you bring clashes with federal copyright law, as Sinatra found out.

Additionally, you have state common law copyright to contend with, as well as First Amendment implications that may protect certain transformative uses of a performer’s right to publicity! Are you getting a headache yet?

In short, recreating the performance styles of famous musicians using this software may not be legal in areas of the law outside of copyright. The same things that hold true for a vocalist’s right to some control over the use of her vocal likeness would hold true for a performer’s right to the use of his musical likeness. In most states, right to publicity extends to the heirs of a deceased celebrity.

If Zenph Studio’s software does what it says it does, it is likely that we will see some kind of legal challenge in the future from one of the performers it adds to its software.

Image courtesy of misterbisson