For my final project for CPSC 183 this semester, I wanted to explore the possibility of creating a website that auto-generates a continuous music mashup from song files downloaded from the Internet.
Before I jump into the details of its implementation though, let’s take a step back for a moment.
First, what is a mashup?
A mashup, loosely defined, is a song that is itself composed of several other songs played simultaneously. The songs are played on top of each other in synchrony so that beats match up across songs and, if done well, the result is extraordinary.
Some mashups are simply two songs cleverly woven together and played all- or almost-all-the-way-through, while others feature hundreds of song clips all interleaved over the course of half and hour or more. It is the latter category that I most enjoy (and that I like to fold my laundry to); I find it fascinating how mashup artists can create an entire musical landscape that shifts dramatically over time through the subtle use of other people’s music.
The end of a good mashup is like the last few pages of a great book, or the final minute of your favorite movie: once you get to it you realize how sad you are to have it all come to an end, knowing that no new exciting plot twists or tempo changes await. So, I thought, what if there was a mashup-generator that created mashups on the fly, generating a completely original mashup that could, in theory, run forever?
The mashup is one of the rare mediums in which the consumer (i.e. the music listener) becomes the creator. Mashups, and remixes of all varieties in the general case, take control away from the original musician and give it to anyone with even a passing interest in the medium and access to a computer. I find it fascinating that modern music mashups are entirely dependent upon modern music-editing software, and yet the creators of Logic Pro or Garageband would never be considered even partial creators of a mashup. This is fair—we are a society that rewards output and not the process leading up to it, and it would in one sense diminish the creative effort of the mashup artist to say his/her creation was partially made by a software developer at Apple. These rules extend to all creative processes, and inherently make sense—should the woodworker have to send royalties to the creator of the cross-cut saw and the lathe? Should Mr. Lathe then give money to the inventor of the knife, or the electrical motor; or to Nikola Tesla, or the discoverer or the electron? Clearly this would be absurd.
But the question then becomes: if I could successfully make this mashup machine (cleverly coined “MashupMachine”), who would own the mashups it produces? True, my website wouldn’t save copies of any mashups produced, but 1) one could still record its output relatively easily, and more importantly 2) even fleeting works of art have owners and creators. It’s not as if a sand mandala wasn’t created by a monk just because it won’t last forever.
So is all MashupMachine output inherently my creative work? On the one hand, it’s just a piece of software, a tool, akin to a Garageband. I can certainly be credited for the tool itself, but not for everything created using it. But at the same time, the difference between this program and other tools is this would be completely uninteractive (at least in the initial implementation)—just sit back and listen. In that sense, the situation resounds with the ethical dilemmas of artificial intelligence—if researchers use a genetic algorithm to permutate pieces of computer code until one code produces a new mathematical discovery, most people would say the researchers get the credit. But once you put a face and human-like limbs on the program and have it walk around like in an Asimov novel, I think most people would say the robot gets credit. Is autonomy a factor here? Does the robot need to be able to solve the problem entirely on its own? In that case, what if MashupMachine had vague controls for things like tempo shifts or incorporated song genres?
I think the real dividing line is sentience—the difference between the “weak” artificial intelligence of the genetic algorithm on its own and the “strong” artificial intelligence of Sonny. But I don’t think it’s out of any sort of universal human understanding so much as it is one of practicality: if the creator of a work is sentient enough to claim ownership, he/she does. If not, someone else does because our laws demand that everything have an owner.
As we become more and more dependent upon machines and automated processes for everything we do, we increasingly face this question of who is ultimately in control. Sure, Isaac Asimov and Francesca Coppa argue that we’re moving towards a future dominated by specialist producers of “content,” be it music or software or education. And yet what happens when these specialists are resigned to the consumer role, when all content is created outside of human control? Clearly this is an extreme situation not in the near future, but we’re already in a time when automated processes can make inventions and discoveries on their own. I expect in the next few hundred years we’ll see a shifting of the laws governing ownership, patents, and copyrights in one way or another, though it’s unclear where the courts and legislators will rest their gavels and pens.
Will non-human created content belong to a related human? Will it go straight to the public domain? Will it be government property? Will it be something else entirely, not confined to traditional copyright and property laws? We’ll just have to wait and see. In the meantime, why not let computers make us some music?
If you want to know more about how I implemented MashupMachine (to avoid having a gigantic blog post), go here.
Edit: you can see the website here. It definitely works in Chrome but possibly not other browsers, and be patient… it can take a little while to start. Enjoy the cacophony!
Yale Law Tech 