March 23, 2019

Music AI systems may take over the music creation process from musicians

music AI system computer

Music truly has a way of unlocking our senses and deeper selves. From hyping you up at a rave and gym to helping you study and focus during work, it plays a key role in the everyday lives of many. It’s also a USD 130 billion industry, depending on which sources you rely on. However, music is about to be turned upside down by, you guessed it, AI systems.

Music in the age of AI systems

Wait, music is core to the human experience: how could a machine replace us? We invented it to begin with, right? Unfortunately, we’re already past that point. Ed Newton-Rex founded Jukedex, a musical AI designed to create instrumentals for various end goals. Looking for a new movie score? Look no further. The system is incredibly prolific: it cranked out over 1 million (not a typo) songs since its inception and shows no signs of stopping. New brands have joined since then, likely hoping to get a share of the background music industry which hit USD 660 million in 2017. As systems continue to improve, it’s only a matter of time until we start bumping AI-created music. If that sounds far fetched, Warner Music Group just signed Edel which creates custom mood-boosting sounds. The future is now (old man).

A cause for concern?

It’s no secret that AI systems might take everything over. They’re already terrifying governments with mass unemployment, to the point where we are now seriously considering Universal Basic Income (UBI). AI systems can already make art, so these further developments should not come as a shock. However, all this begs a much deeper and tougher question: are seemingly “creative” fields no longer a haven for humans? Machines already outperform us at a myriad of tasks, but surely they cannot take away what makes us truly unique: our ability to think and execute broadly. In many ways, time will tell just how far these systems will evolve. We’d venture to say that it’ll be far beyond what we imagine.

Taking a step back

Perhaps we assume that our innate creative abilities are beyond where they actually are. If we stick to pop music, many of your favorite hit songs were made using just four chords. The Axis of Awesome brilliantly illustrated this a few years ago with their now-viral video. This should not be surprising: music relies on simple and subtle mathematical formulae that best please our brains. The Greek philosopher Pythagoras (the one with that pesky theorem you couldn’t remember in high school) figured this out a long time ago. Turns out it doesn’t take all that much to please our brains with a nice tune. Given how seemingly simple it all feels, it may be no surprise that we can code this into a neat algorithm altogether. Should this put into question our core creative skills? Not all craft is the same, which is probably why we don’t compare Mozart to Nickleback.

Human history: a silver lining

It pays to know how machine learning systems work. To over-simplify, these programs rely on massive datasets to draw conclusions and hone their skills. Jukedex is the next iteration and uses deep-learning instead. However, music datasets require existing musicians’ work, begging new questions around copyright and royalties. If scientists feed “Crank Dat” into an AI system, Soulja Boy should get rewarded for his contribution.  Therefore, every prior musical piece ever created may play a pivotal role in these system’s development, opening new potential avenues of monetization for artists. More importantly, these systems exist because of human contributions, not the other way around. Humans should feel deep pride in their ability to generate meaningful things over generations. Music continues to stand the test of time, often blooming in our darkest times.

The larger picture

If we’re honest, AI systems will make great music within a few short years, but musicians should not worry altogether. Often times, music relies on great personalities that drive cultures forward. No computer can replace a human being, even though we do love our digital lives and identities. Even if it’s a hologram, seeing Tupac owning a stage feels more meaningful than watching HAL 9000 do the same. Tupac’s music tells a larger narrative of Tupac’s life and his environment (which people can latch on from a storytelling standpoint). Even if a computer were to write relatable lyrics, audiences would need to somehow empathize with a cold machine’s teenage angst. This seems like a stretch at best, but not entirely out of the question. Creatives should also rejoice at the idea that computers can speed up their workflow without impeding on their style. Computers can easily create, but they cannot easily connect. Time will tell if this continues.

House music will never die

Computerized music isn’t anything new. From synth tracks to Daft Punk and Kraftwerk, man and machine have worked intimately together for years. Your favorite trap beats were made on a really crummy sounding drum machine (yes, they meant to imitate those drums). In hindsight, this new wave of machines should not scare us, but rather give us hope to create better and deeper music. Tools remain just that: tools to improve a workflow. Humans will ultimately validate what is “good” compared to what is “bad.” If man and machine already work seamlessly, then (house) music will never die.

Look forward to more great tunes in the future. You can even go buy or steam our very own Delf’s new tracks here and unwind this weekend and enjoy it with beautiful imagery here.

Play Pause