A Robot Has A Record Deal: How Artificial Intelligence Could Save Music
Warner music group just signed a deal for 20 records made by an algorithm.
Despite movies like The Matrix trying to warn us against the dangers of artificial intelligence, it appears humankind is walking eyes-wide-open into an AI induced apocalypse.
Look around, signs that malevolent technology is already among us are everywhere. Deep down, you know Siri resents you, you can hear it in her tone when you ask her to place your saved Domino's order. Your smart TV didn't glitch; it deleted your taped episodes of The Bachelor out of spite. Alexa didn't mishear you, she's just tired of your shit.
Now, you can add music to the ever-growing list of things AI will inevitably use against the human race when our iPhones enslave us. For the first time ever, an algorithm has landed a record deal. According to Music Business Worldwide, "A stress-reducing sound app called Endel has become the first-ever algorithm to sign a deal with a major label, Warner Music Group, and will be releasing 20 albums throughout the year."
This doesn't exactly mean Plankton's wife, Karen, will be the next big American pop star, as Endel is in the business of "relaxing soundscapes," not mainstream music. The app uses a variety of factors such as time of day, location, heart rate, and weather to create custom sound frequencies to relax the listener. While the existential dread we feel when listening to music written by a robot is anything but relaxing, we're sure some people can look past the spookiness to achieve a moment of big-business-manufactured zen.
Still, if algorithms can play you what you want to hear when it comes to a mixture of rainforest sounds and white noise, what's stopping them from creating the next perfect dance hit? Nothing. In fact, it's already happening. While this is the first time an algorithm has ever had its very own record deal, producers have used algorithm technology in music production for a long time. And maybe, it's not quite as scary as it sounds.
According toThe Verge, this technology works by "using deep learning networks, a type of AI that's reliant on analyzing large amounts of data. You feed the software tons of source material, from dance hits to disco classics, which it then analyzes to find patterns. It picks up on things like chords, tempo, length, and how notes relate to one another, learning from all the input so it can write its own melodies. There are differences between platforms: some deliver MIDI while others deliver audio. Some learn purely by examining data, while others rely on hard-coded rules based on musical theory to guide their output."
If you listen to the songs that have experimentally been produced by AI, it quickly becomes apparent that something is missing. "Daddy's Car," an AI made pop song created under the supervision of songwriter and human man, Benoît Carré, was made by Sony's Flow Machine software, a program that analyzes a database of some 13,000 tracks from around the world to create music.
Daddy's Car: a song composed by Artificial Intelligence - in the style of the Beatlesyoutu.be
Carré taught the software to use the music of The Beatles as a template, and while that influence is apparent, something is hauntingly off. It's tempting to say that the song is missing a vital humanity; a warmth, intention, and spontaneity that the human mind connects to subconsciously when experiencing a piece of art. But in all likelihood, the problem with "Daddy's Car" is simple: it's just not a very good song.
As Hang Chu, a computer science Ph.D. student at the University of Toronto explained, making music with AI is different than, for example, generating images with AI (something that's been done successfully and convincingly for a long time). "Music is not something where if you throw enough data at it and hope the algorithm can figure it out, it will work," he said. Considering the number of factors that go into making a piece of music: tempo, tone, melody, harmony, and timing, this makes sense. Even though these factors are quantifiable, it's a shot in the dark to know what combination of them is going to create a good song. After all, if making good music was an easy math problem, everyone would do it.
Musical AI still needs a person to operate it, and ultimately, make computer generated products into a song that a person would think is worth listening to. There is still human artistry in AI music technology, just a new and different kind. While most people react with anger and discomfort when confronted with the idea of computer-made music — immediately jumping to arguments about music being at the very core of what makes us people, so should, therefore, be made by people — isn't it true that the technology of music has always been evolving? Minstrels didn't have electric guitars and George Martin didn't have Pro Tools; how is AI anything but a new man-made instrument?
John Smith, a fellow at IBM's AI research center, argues that AI-made music may actually help to take music to places it could never get otherwise. He says that we need to widen our ideas of "good" or "bad" music or art when it's created by AI. "Part of the joy of AI is that it doesn't think like humans do, and so it comes up with concepts and ideas we would never think of. That can be helpful for pushing art into new realms," Smith said, but added, "The computer can start to do more and more of the groundwork and prep work and even suggest different ideas, but that leap of creative thought, that spark of imagination, still has to come from a human."
Not only that, but your taste in music is already probably heavily influenced by computers. Algorithms created by streaming platforms determine what kind of songs they think listeners are going to like, and then push these songs to listeners, inevitably boosting the popularity of certain kinds of music, which serves to influence musicians into copying this sound. People readily accept this kind of computer interference — interference that is arguably more damaging to the autonomy of the music listener than AI music.
So, though music generation software hasn't quite learned its trade well enough to create something convincingly human, inevitably, it will. Luxembourg company, AIVA (Artificial Intelligence Virtual Artist), created an AI-generated score for a video game called Pixelfield using a slightly different method than Sony, taking into account that the specific combination of different musical aspects, not each aspect on its own, is what makes a quality piece of music: "We asked, 'what are the building blocks to create an entire song?'" Pierre Barreau, the CEO of the company said. "If you consider just melody, you can generate that. Then based on that melody, you can make another model that creates instrumental accompaniment for that melody. If you break it down, it becomes substantially easier."
It's naive to believe that as new systems like AIVA's are invented, the technology inevitably improves, and AI-made music becomes something you hear more of, there will be an inherent lack of humanity. After all, it's human created software, run by a human, using building blocks made of patterns and aspects of songs produced by people; how can the resulting whole not be imbued with the same humanity?
- Amper Music: AI Music Composition Tools for Content Creators ›
- Google A.I. just created music (CNET Update) - YouTube ›
- Jukedeck - Create unique, royalty-free AI music for your videos ›
- Daddy's Car: a song composed by Artificial Intelligence - in the style ... ›
- The World's First Album Composed and Produced by an AI Has ... ›
- How AI-generated music is changing the way hits are made - The ... ›