The music industry has, historically, had an ambiguous relationship with technology.
Ultimately, though, the industry has ended up capitalizing on the very technology the major players wanted to shut down, the gramophone, the cassette, the CD etc.
Innovation in music technology has always equaled new business opportunities and growth in the music industry. That is, until «the MP3 revolution» kicked off a new era in music distribution and a redefinition of the core artifact/product.
Today, this is a familiar story. With the advent of «pirate companies» like Napster, Limewire, etc. mp3 versions of songs were available to download on your personal computer for free. Simultaneously the film industry had started experimenting with streaming technology.
The music industry never put two and two together, recognizing that the home computer/laptop, mobile etc. would be the new distribution and consumption channels for music, and that streaming technology would be the solution.
Litigations became a key strategy, combined with implementing copy-controlled CDs that, in many cases, were rendered useless on several devices. The lawsuits as well as the «copy control» tech alienated the consumer.
This (lack of) strategy led to letting tech companies take the lead, and we got Spotify, Tidal (originally Wimp), Deezer etc. in addition to Apple Music, Amazon Music, YouTube Music etc.etc. Failing to identify the momentous opportunity in digital music distribution, the industry relinquished control of one of their key assets, the album.
While major record labels have never been known for their altruistic approach and generosity towards artists, in a «new reality» with tech companies dictating the terms, the artists ended up as «the biggest loser». And, there is a sad irony in this that music consumption is higher than ever, whereas earning a living as an artist is probably harder than ever (since the industrial revolution).
Now, as the music industry stands on the cusp of what is likely to be the biggest revolution ever, there are numerous new challenges, and maybe opportunities too, on the horizon.
Generative AI starts making a noise
The first AI music generators for the general public popped up with Jukedeck a website launched in 2015 (acquired by ByteDance in 2019).
Although Gen-AI has been around since since the 1960s most of us were blissfully unaware of the capacity of the technology until 2023. Everyone started paying attention when ChatGPT was launched.
Soon AI music generators like Suni and Udio followed, allowing users to «create» music from prompts. These models are naturally trained on existing music, and generate new versions of what a song in a specific genre sounds like.
This is not without complications as the output will inevitably be too close for comfort to existing copyrighted material. And, the major labels are, unsurprisingly, suing the companies behind these services.
Universal, Warner, and Sony are suing Suni and Udio for «using copyrighted sounds and songs to train the artificial intelligence that powers their businesses.» The AI-generated tracks are like a mega «remix» of hundreds of thousands of songs, and it was just a matter of time before copyright holders would react.
The quality of the services is rapidly improving, and will soon be indistinguishable from human-made songs, for the general listener. At this stage, there will be a radical paradigm shift in the relationship artist/listener.
Music on the Grid
Why are we seemingly so accepting of «machine-made» music? I believe it has to do with how we are used to consuming music as well as how it is produced and distributed.
Click tracks, quantized beats, and auto-tuned vocals are the tools of the trade when producing music. Later the finished product is distributed to platforms like Spotify that offer «high-quality» streaming at 320 kbps, whereas the .wav file format that studios normally use is at 11411 kbps.
Basically, with Spotify you lose 3/4 of the original audio data. And, finally many listen to their playlists on subpar PC speakers or headphones or a cheap Bluetooth speaker. Ultimately, this is not a value chain that truly values organic, high-quality performance and audio.
With all the tools at an artist’s disposal today to perfectly place their music on the «grid» (of rhythm and harmony), there is no challenge in playing «precisely» or «in tune». This is all easily fixed by the software. Artists stand out by having truly unique ideas and artistic personalities and having the courage to let these reverberate in sound.
This is not a plea to abandon digital tools, but to take the creative, rather than the lazy route with these. One example of an organic approach is, what is arguably the most influential of all HipHop producers, J Dilla.
When Dilla was active (1995-2005), it was common practice to compose beats on a digital pad representing various percussive sounds, using your fingers. By refusing to quantize (the software perfectly places the notes on the «correct» rhythmical grid) the beats, he created what is referred to, with reverence, in the world of HipHop, as «Dilla Time». This is a groove where the rhythmic feel is completely organic, notes are «stretched» and «shrunk» and fall «off the grid».
In contrast to Dilla’s organic approach, we find basically every single major pop artist today. Everything is produced «on the grid», and compressed for maximum loudness on digital platforms. Furthermore, the actual song composition adds to the «impersonality» of the product with song credits reading like the line-up of a football team. This is due to the practice of having specialized writers for every single part of a song (beats, hooks, lyrics, harmony etc.etc.), to make sure the product is streamlined to the max.
How about the singular songwriter, then? Considering that Taylor Swift is using the same four chords for 90% of her hits and Ed Sheehan has made a career of «repurposing» other song material for his songs, they are arguably functioning quite like AI music generators already!
This streamlining of the whole process of creating music is locking the listener into a «mega-grid» of an exceedingly consolidated commodity.
Finally, a couple of words regarding auto-tune. This is used all across the board, sparingly for pop artists who «can», in fact, sing, and full-tilt for artists who cannot, in genres where the tool is part of the «sound». Acclaimed UK pop artist Charlie XCX recently stated in an interview, “I could sing in tune but now I think I’ve gotten so lazy because I sing with auto-tune all the time”.
And, why should an artist work hard on having good pitch, and precise phrasing when the producer can easily fix it in the mix? There is a beautiful tension in struggling with imperfection that is lost when there is no struggle anymore.
How the listener has had the auditory process «sanitized» by auto-tune is an even bigger threat to genuine artistic expression. Being completely immersed in voices that have been digitally processed lays the ground for a quick adaptation of 100% ai-generated voices. It is already becoming difficult to identify what voices are real or AI.
General AI and Music
What is next on the horizon?
Artificial General Intelligence (AGI) or General AI represents a theoretical form of AI that can solve any task using human-like cognitive abilities. This means that AI can break the shackles of specialization and learn and apply new knowledge across a vast spectrum of domains. Predictions vary greatly as to when AGI will be a reality, but many believe that it’ll be here by 2030.
General AI will be able to switch between recognizing speech to performing numerous tasks through any discipline and even understanding abstract concepts. At which stage, autonomous music composition is not beyond its capabilities.
Future developments in AGI might involve other techniques than Machine Learning like evolutionary algorithms or symbolic reasoning, potentially leading to even more obtuse and complex copyright issues for the music industry.
There might come a time, in the near future, when the music industry will look back at this stage of the development with a certain wistfulness, a time when gen-AI was using ML models, and it was fairly easy to claim ownership of sound.
What can labels do?
It is unlikely that one will be allowed to AI generate music and then release it to the general public. However, if users are happy with generating AI songs for their daily soundtrack, then that represents a gigantic revenue loss for the record companies.
One obvious solution is a «licensing fee» with a revenue share agreement between AI music generators and labels. Interestingly, the AI companies, like the «music pirates» of old decided to circumvent completely the collaborative approach and decided to «go their own way». This has opened them up to lawsuits, as we’ve discussed, but also to form «allegiances» directly with the consumer, outside of the regular channels.
In another predictable move, major labels are attempting to wrest control over AI music generation by offering proprietary services. Universal Music Group (UMG) is partnering with SoundLabs which has developed MicDrop, a real-time AI vocal plug-in that allows artists to create high-fidelity vocal models using their own voice data.
“This will allow UMG artists to push creative boundaries using voice-to-voice AI to sing in languages they don’t speak, perform duets with their younger selves, restore imperfect vocal recordings, and more.» (Chris Horton, UMG)
I question whether UMG’s approach to AI music generation is «good for music», or artists, but it might turn out to be «good for revenue». This will, of course, open up new revenue streams for the artist, but the question is how it will affect the listener re/inter-action with the «real music» of the artist in question.
A novel approach to the challenge of AI music generators is to take steps to reevaluate the value of artistic creation. This unlikely strategy would entail to resist the temptation to constantly seek out new ways to embed AI in music making, and to invest in unique artists and give them the budgets to create real-sounding music.
This doesn’t mean ignoring the challenge of AI music generators etc., but using its «creative productivity» characteristics to enhance the artistic endeavor without interfering with the real artistic expression. I feel what SoundLabs is offering is, more than anything, artistic laziness (as described by Charli XCX previously). And, this is not a long-term viable strategy for the industry.
What can artists do?
-
- Be as authentic as possible. In stead of trying to conform to «the Grid», embrace all the idiosyncrasies of your artistic persona. Standing out when 100K songs are published on music streaming services daily is a momentous task, but it is easier if you don’t turn down the weird and unique frequencies of your sound.
- Engage with your fans. It is not exactly a new insight I am sharing here, but now more than ever, this is a key career survival task. Some artists are actively interacting on SoMe, others communicate via their blog (like Nick Cave) or even arrange events with fans. You need to answer the question, why should anyone care about my music? They will care if they can relate to you as a person.
- Make the machines work for you. It is not my intention to be a Luddite and anti-AI. In fact, here at notefornote, we work to develop solutions for music promotion, using Machine Learning models and more. AI can be a great partner in producing art, but my advice would be to its use to solve productivity, not creativity.
- Take control of your career. Whether you are on a label or not, being informed about all the stages of bringing your idea from your head to a market is essential. Most artists do not want to be professional promoters, managers etc. but taking informed decisions regarding how to present and promote your music is essential. In fact, artists should also be in control of their posthumous careers to avoid that their likenesses and voices are being (mis)used after their passing!
In conclusion, the solution for labels and artists and labels alike is simple, take a bet on human expression and take advantage of Artificial Intelligence to produce and promote this expression.
Freddy Nietzsche (not to be confused with Neil Young producer, Jack) coined the phrase «Human, All too human» as a central concept of his philosophy. He challenged us to confront human nature’s complexities and contradictions and embrace our human limitations. This is the way to overcome our flaws and frailties and to become, a true artist.
