Artificial intelligence (AI) and music are intimately linked, but will AI help artists create new musical genres or replace musicians altogether? AI is the question explored by Boris H, reader of Usbek & Rica, with the dual role of artist-musician and developer.
The evolution of music has always followed the development of technology. It is the first area to have been affected by technological advancements. And it is for this purpose that music is inseparable from technology.
Music and technology have always been intrinsically linked with each other. Technical upheavals have changed the way we think about music and the way we listen to it.
Thus, faced with the next technical advances with AI emerging, the musical field is still in danger of changing. Technology will push the boundaries of creativity even further, and new art forms will emerge.
Musical AIs are already here.
AIs are seriously starting to invade sound waves. The AI named Juckedeck has already composed more than a million music pieces for commercial purposes. Over the last few years, several similar firms like Amper, Popgun, and AIVA have emerged to join this new and strange industry.
Their tools are simple with one click: choose a genre, mood, duration, and boom, AIVA or Jukedeck generates a free composition for your project or, if you pay it, for commercial use.
Songs composed by AIVA and its ilk are already appearing in podcasts, video games, and YouTube content – this can range from background music for family vacation videos to sports videos.
For years, Sunday videographers had to retrieve music from huge music libraries, whose works were produced by humans. The AI now offers more or less personalized compositions with just the push of a button.
Sometimes the songs can be surprisingly good. For fun, I created modern cinematic music on AIVA by choosing the Epic Orchestra option, with violins, double basses, and percussion. The AI even added a few small melodic breaks. As a musician, I was impressed by the overall consistency of structure and harmonies.
This tune was neither brilliant nor memorable, but it quickly matched the quality of human work you hear in some videos or commercials. No one could tell the distinction between music composed by humans.
To achieve the same result, it would take a minimum of ten hours, even twenty hours, to compose such a human composer piece. AIVA did it in less than 3 minutes. For example, I work between 30 and 80 hours to make a single piece of music.
However, AIs can’t do all styles yet. They cannot yet handle an electronic piece’s complexity, which requires many more parameters than the elevator music they currently produce.
Artificial neural networks require thousands of examples to understand how to make a melody or rhythm. If you feed AI MIDI files containing Bach or Beethoven melodies, they will understand the mechanics and intricacies and then play the same patterns.
AI knows how the notes come to sound together. The more the AI is fed with good quality examples, the more it will be useful musically. Of course, this means that AI needs to guide in its learning. She won’t learn to compose like Bach by magic if you don’t give her examples from this artist.
AI without emotion or experience
It’s almost systematic. Whenever we think about the impacts of automation, there are negative and positive prophecies. Optimists say AI will destroy some jobs but create new ones that pay better and require more creative intelligence.
The pessimists reply that they agree that these jobs will never be numerous enough to employ everyone or not materialize quickly enough. Entrepreneurs who are behind creative weak AIs usually fall into the first camp.
Their efforts risk destroying all prospects for novice composers without virtuosity, but they will not eliminate the need for talent.
Indeed, people who write complex sheet music for film, television, video games, and music that we want to listen to for fun will still be in demand for some time. No current system is close to human creativity today. And the goal of these companies is not to replace musicians. It wouldn’t be a good thing for musicians.
These algorithms cannot yet understand the context or purpose. It’s like having a composer who has no other life experience than reading 10,000 pieces of music and then trying to do something similar. Therefore, he will not know what he is doing and will not bring any emotion or experience to life. Creative weak AIs have no idea what is culturally relevant, useful politically, or anything else.
In a sense, neural networks only simulate human makeup. We, too, consume hundreds or thousands of music in our lifetime, which gives us intuitive patterns and then recombine that knowledge into something new.
We sample, we fly, and we modify. Our creativity is based on that of those who came before us. But when a machine does it, it can feel like an impersonal, even vampiric act. As of yet, no AI is good enough to create, on its own, a whole decent symphony or even an entire pop song with lyrics (some artist-programmers have tried to generate lyrics with limited success).
Even though an AI from Huawei was successful in 2019 to complete Shubert’s unfinished symphony, she couldn’t make one on her own with a real musical personality. So if we want to schematize the El Dorado of the collaboration between human creativity and artificial creativity, it would be this: a hit song in one click. The last fight of humanity!
AI, a tool for a new musical genre?
AI promises to democratize composition in the same way that Instagram makes each of us a photographer. Artificial Intelligence will undoubtedly soon integrate into instruments, into real hardware shortly. AI will be in keyboards, pianos, guitars, basses, and it will be everywhere.
Imagine sitting down in front of an electric piano to be guided by Bach’s ghost. This will probably sound funny to you, although it also raises a disturbing possibility: Won’t such instruments start to unsettle young musicians and make them decide not to continue mastering them?
Rise Of Artificial Intelligence AI
On the other hand, the rise of AI could spawn whole new musical genres. New technologies often do. The electric guitar gave us rock; the synth contributed to the New Wave, electronic drum machines, and samplers to hip-hop and electronic music. Autotune started as a little secret in the recording industry, which helped clean out wrong notes until artists like T-Pain used it to create new vocal styles.
There are still challenges to guarantee artificial intelligence that is ethical, responsible, and creating value. The challenges will not only be for humans; machines too will have to deal with many challenges: working on small tasks, interactions with the physical world, common sense, knowing how to explain what they do, why and how, giving their opinion, understanding routines and human emotions and their multiple interpretations.
The AI will not replace the musician right away. On the other hand, it is already creating opportunities and will democratize access to music.