The Impending “Death” of Music… And Why That’s Fine
In a world where technology is increasingly fulfilling your wildest dreams and desires, there’s bound to be points where technology veers into uncanny areas. Take machine learning for instance, where a computer can analyze data with a layered structure of algorithms. As a matter of fact, it is very similar to the way a human might learn. Sounds cool, until you realize that means that computers are becoming humanlike, and will very well take our places in society. You’ve already seen robots that manufacture car parts, reducing the need for low skilled workers, or maybe even algorithms that can control the military without human intervention. But is it really possible for computers to replace musicians? Music is a form of art, so how can it possibly be made by robots?
AI is taking the place of humans in the process of making music.
It sounds cool, it sounds gimmicky, and to musicians, it sounds scary.
This idea has been teased since 1960. That year, Russian researcher R. Kh. Zaripov published the first paper on algorithmic music composition by computers. In 1965, the idea was actually executed by inventor Ray Kurzweil, who performed an original computer-generated work of classical music on live TV.
Half a century later, programs have been developed that can generate all different genres of music with different instruments and can create catchy melodies. They all use machine learning, so all you need to do to use it is input a number of songs, let the program analyze them, and then the program can output an original song similar to the songs you gave it.
An example of such is EMI (a computer program, not the record label), which has been developed by David Cope since the 1980s. It specializes in generating convincing classical music. In 1993, Cope actually released a collection of Bach imitations created by EMI, nicely titled Bach by Design.
Another example is Humtap, which can use machine learning to pinpoint the musical workings of a band. They boast on their website: “For consumers, we can make new music with an artist in mind with razor sharp precision. For artists, we they can create and generate new material of our own in no time.” The only problem with that statement is that the first point counteracts the need for the second point. Why would a consumer buy or even listen to music by other artists if they can just as easily listen to music tailor-made to their preferences?
Definitely, the most high-profile AI-involved projects have been within the last couple years, namely I AM AI by Taryn Southern and Hello World by Skygge. For both, the lyrics and voices are the only things human-made. Everything from the melodies, chord structures, instrumentalizations, and even sound mixing were automated with a computer program. And the best part: they’re both pretty catchy.
Another band that uses AI is DADABOTS, who performs all their own music, but uses AI to generate interesting heavy metal compositions.
Record labels might already be rushing to get their hands on programs that can automatically generate chart-topping hits. After all, having computers write high-quality music would be cheaper and faster than forking over millions of dollars to artists. And even after that, record labels might become extinct once consumers get their hands on programs that can just endlessly create music to their own likings. As far as the artistry goes, AI is going to truly be the death of music.
It’s sad when you put it that way. But think about the possibilities of the perfect computer that can make perfect music. Realistically, it would be able to make songs of a higher caliber than any human could. Because of the restrictions of machine learning, it wouldn’t be able to explore new types of music, but that’s the only restriction. AI could open the gates to perfect songs, ones that everyone could enjoy and play 1000 times without getting tired. Songs that you can have stuck in your head for life. In fact, AI will make these songs. If you are a composer or songwriter, it makes sense to be bitter about this, because you will be out of a job when the AI music revolution happens. It’s like how factory workers are being replaced by robotic arms. My advice: don’t pursue such a profession in the near future.
However, for the rest of us, this movement is something to be hyped about. I see AI like monkeys on a typewriter. Just like how the monkeys will eventually type out Shakespeare from randomly pounding on letters, AI will produce the one perfect song that is best song you will ever hear. Except AI can do it in under a second.