Music’s Midjourney Moment
Music’s Midjourney Moment
The Future of Human Music and Art is Being Disrupted
It’s crazy to think that its been well over a year since the generative AI image service MidJourney bubbled up to the front of public perception.
At its start, it was novel at best, spitting out deformed shapes and compositions that weren’t necessarily beautiful to look at, but impressive nonetheless given they were being drawn, like magic, from the digital ether through the power of machine learning and AI.
Fast forward to the present and with a few sentences of text description you can output anything from photo-realistic masterpieces to the wildest avant-garde imagery that your imagination can conjure.
Once it hit that tipping point at the critical mass of quality, everyone was exposed and now generative imagery has become part of our modern digital media workflows.
Music is following a delayed, but similar journey. It is just about to hit that same moment of impressive disbelief that MidJourney did right before AI images exploded into the mainstream.
AI Music: Slowly then All at Once
I’ve been following AI music for years, mostly keeping an eye on Google’s Magenta project which has had impressive, but fairly rudimentary breakthroughs along the way. I’m not a machine learning expert, but I learned a good amount from following that project.
As I understand it Google, and subsequently the Magenta project were the first to author and apply the “transformer” deep learning architecture that is now the basis of all the major AI advancements that we are seeing on the consumer side, including ChatGPT.
During the time I was following their project, it took a decent level of technical knowledge to get anything interesting to work.
And for a while I figured there would be a whole, fairly long era of nerdy musicians piecing together code to produce AI generated parts and tracks.
But then all of a sudden, all at once, fully produced tracks are now available to the public with only the entry of a short text prompt.
I am continually blown away by the speed of advancement of AI and how fast everything is moving. It is truly exponential growth as new advancements are ushered forth faster and faster based on previous gains.
The interesting thing about this era and the speed of adoption of new technology is that it has never happened this fast before.
Humans are creatures of habit and we take time to adapt. There have been many in our history, the printing press, the combustion engine, the internet, the iPhone, but with each advancement the cycle shortens.
Now with AI it’s faster than ever.
I digress, and maybe that’s a broader topic for another entry, but with music the new capabilities that are emerging are far from what we’ve seen so far.
A New Level with Udio and Suno AI
In the past month or so, two leading AI music services have released new versions of their text to music generation models that truly demonstrate the speed of these advancements.
Just a couple weeks ago former researchers from the Google Deepmind project launched their new generative music app, Udio, which can generate a short song, complete with vocals, in just about any genre, existing or completely new, that you can imagine.
Rather than attempt to describe in words what it is capable of, here are some examples.
Dune Broadway - https://www.udio.com/songs/eY7xtug1dV6hbfCDhyHJua
Pop Punk (complete with annoying vocal) - https://www.udio.com/songs/4vHynYzJpxULVB1PNzvuj7
Heavy Rock - one I created about how synthetic singers have higher range (not nearly as impressive though, right?!) - https://www.udio.com/songs/cCYg39Sc1g1uwxo5AtBQvC
Another extremely impressive music generation app leading the charge is Suno AI, which announced their public release not too long before Udio.
Their app doesn’t seem to have quite the polish as Udio in its current version, especially with vocals, but it demonstrates some extremely impressive results as well.
Here is one that includes a rad 80s metal guitar solo:
Hair Metal (ripping guitar solo) - https://suno.com/song/02504260-eaf3-4e21-ad20-cdff56791186
I knew that AI music generation would be at this level eventually, but actually hearing it is pretty mind blowing. You can imagine where this technology will be in another year or two. It’s crazy to think about.
As someone who has dedicated at least several thousand or more hours of his life learning the skills to write, produce and perform music that AI can now do within seconds, I definitely have questions, as I’m sure you do as well.
Casual Song-prompting (Songwriting?)
It’s interesting to me how this technology is rolling out, but as I think about it, it makes sense from a business perspective.
As I followed Google’s Magenta and other AI music developments over the past years, I assumed that the tools and capabilities that came would come available first would be more as part of music production process, helping writers and producers with individual parts of the process such as mixing, midi generation, drum fill generators, etc.
But now it’s clear that the biggest impact on the market will be the AI generation of entire tracks of music that completely replaces the existing music production process.
The biggest reason for that is: business. And to a secondary degree, marketing.
Music writing, production and performance have been, and still are, pretty difficult processes that require thousands of hours of investment in knowledge and practice to become good enough at to create releasable tracks that would be appealing enough for people to want to listen to.
Because of that, the amount of people that would create music in the past is a fairly small part of the populous. In a business sense, a small market to which to sell products such as instruments, mixing tools and software, and other services that are a part of the traditional music making process.
But almost EVERYONE wants to create music, even if most aren’t willing to invest the hours to do so in the traditional way, even with recent advances such as digital audio workstations in the recent years.
This opens up a huge market with AI for music fans to now create their own music as simply as entering a text prompt.
With an expanded market of exponential degree over traditional music makers, the profit possibility is enormous. It is essentially a new and untapped market and with it the possibility to make a lot of money.
For this reason, the purely generated text-to-music side, with its ease of use, is what we are seeing first in the market. It provides the greatest potential for growth and profit.
That being said, I think the advanced AI tools are still coming to serve the traditional music makers, and that is the part I’m actually pretty excited about.
Tools that help you focus more on the creative side of creation rather than the technical aspects of mixing and sound design. Tools that enable a new level of possibility for musicians, producers and composers with talent and vision.
The Copyright Issue
This topic is a bit outside the scope of this article, but worth mentioning. There is a big argument and conflict arising over how AI models, used by apps such as Udio and Suno, are trained and whether or not they use copyrighted work without permission.
There is more and more evidence arising to support that they do, which brings up all kinds of new arguments about intellectual property and copyright in the AI era.
There are two camps on this topic in the music market, basically the e/acc (Effective Accelerationism; basically tech advancement by any cost or method) and, traditional artists.
AI, and how it is created / trained and used in the open market, is a new area of commerce and with it comes the discussion of new laws and definitions around copyright.
Using copyrighted material without permission is easy to be viewed as not a good thing, but knowing the tech startup mantra, I believe these companies are racing to push the needle far enough fast enough that argument for advancement will compromise traditional copyright protections.
Personally, I see both sides. I’ve devoted many thousands of hours into the skills required to write, perform and produce music and it pains me more than a bit to see and feel that devalued, but I’m also a technologist and believe that progression and advancement are always the way forward, if done so ethically and compassionately.
The truth is, disruption and advancement have always been a part of our society and economy and it has often caused some pain in the more drastic instances.
AI is unique in this though, I believe, because of its exponential ability to scale.
Some more rudimentary arguments on the e/acc, tech forward at all costs side say that music should be a hobby and passion for creating should be devoid of the need for money, but the existing many billion dollar industry would probably disagree.
The truth is, money is a tool and resource that fuels the energy and time needed to create anything, whether a business, or art, or the business of art. The more money you can make off your art, the more time you can free up by not needing to work other jobs, and subsequently more you can focus on mastering your business or art to become something truly unique and valuable to society.
That’s how economy works and how we’ve gotten so much valuable art and other ideas to come to exist in society.
Again, a broader topic for further discussion in the future, but something that is rapidly developing and affecting the future of the industry and existing traditional creators.
What Will Happen To The Music Industry?
It’s pretty clear that the emergence of AI generated music and the quality it is already exhibiting will have a big effect on the music industry as we know it.
Outside of pop stars and rockstars there has long been segments of the behind the scenes industry that have been lucrative for music creators that aren’t public artists.
Library and sync music, for instance, have long provided a way for songwriters and producers to make money either full time or on the side as they try to grow as an artist.
Composers creating music for film and TV extend down from the top tier of known names to the secondary ranks of media providing a way to make a living for talented creators in the middle class.
Unfortunately, I believe those areas of the music industry will be exposed to the biggest disruption from AI generated music.
Having dabbled in the past with library music production and composing (which I absolutely love to do) as a means of supporting myself, I know that in the up-and-coming levels of film and media production, budget is always an issue. If producers and directors can get a close enough score or placement for a fraction of the cost of human made music using AI, they will absolutely go with the monetary savings.
That said, from the perspective of the director and film creator, this is a positive thing. There is only so much budget and money that can go around and creation and completion of the higher vision of storytelling is the goal. AI music will allow writers and directors to become composers in their own way with AI doing the heavy lifting of composition, theory and music production.
I strongly believe in the talent of music writing and composing and that will absolutely still exist at the higher levels where there is budget, but the cost saving potential of AI is too great to ignore.
Furthermore, other areas of “functional” music, like mediation, yoga, and commercial scores and licensed music will very clearly be the work of AI in the future, at least in the majority.
These changes in the balance of creation and industry can certainly be seen as both a positive or a negative depending on your current position and view, but the fact is significant disruption is happening and is coming fast.
On the other side of the industry involving labels and artists, I definitely see the trend of pop factory, formulaic pop star creation continuing to an even greater degree as labels get hold of AI that can mix and match hits from past decades into “fresh” new songs and sounds that have the built in assurance of investment of past hits (cue again the copyright and training discussion).
Again, the trend feels like its been going this way for awhile, but I see the divergence of “entertainment” and “art” separating even further as AI music potential grows and the label money side of the industry pushes for the “sure thing” regarding their investment.
To be honest, I don’t intend to sound negative with any of this, disruption is change, and sometimes change can be uncomfortable.
At the end of the day, music is art and art is expression and connection between humans. And humans as a collective define demand and value within a market.
The further shift to entertainment vs art by the traditional music industry and its investors has been happening for decades. And I personally feel this is creating a hole where people are more and more creating connection and authenticity.
AI tools and capabilities will also provide independent creators to do much more with less and start a movement to empower communities around authentic art devoid of profit hungry gate keepers.
The Future of Human Artists
As AI music develops, I think the music industry will segment into 3 different categories:
Label Driven Entertainment
Consumer Created Music
Independent Artists
1. Label Entertainment
As we just discussed, over recent decades, music labels have been more and more focused on turning a profit on their investment by sticking with the “sure thing” and not taking too much risk with the artists they sign and support. As AI music grows, I think it will support that framework, allowing them to focus on signing the new pretty face and then using the help of AI to generate music content inline with what has worked so well in the past.
That combined with the ability to leverage the vast catalogues from hugely famous legacy artists to create new value around them. Imagine a new release from huge artists of the past that are no longer with us such as Hendrix, Bowie, Elvis or John Lennon with The Beatles (which has already happened: “Now and Then” with John Lennon)
This legacy capability will create huge value for labels as it taps into nostalgia for personalities and sounds of the biggest artists in history during the era (now fading) of larger than life rockstars and celebrity musicians.
2. Consumer Created Music
AI text-to-music capability as we’re seeing with Udio and Suno will create an entirely new market where music fans and listeners will be able to create their own music by just describing a song with a text input without the need for thousands of hours of practicing guitar or learning to sing in key.
Curiosity will spark endless creation as users follow their prompt curiosity and mix and match genres and remix existing artists tracks with new vocals and sounds.
The tech companies championing this full generation capability of AI music will create and tap into a brand new market of users who love music and want to create but don’t have the skills to do so through traditional methods.
3. Independent Artists
There is a reason that music has become such a big part of our lives and society. It is a part of our essence as humans. It has a spiritual quality. The connection that happens from forming an emotional expression to which we can all relate into a beautiful sonic medium is nothing short of magic and in my opinion is such a beautiful piece of the human condition.
The artists that have the ability to so aptly channel these feelings are truly special and will always supersede the ability of machines and casual creators.
I don’t see this aspect of music and art ever completely going away, and it certainly won’t be replaced by machines.
New advancements in AI and digital technology will enable true artists to channel their visions and forge deeper connections with their audience without the support of labels and the huge budgets needed to do so in the past.
Focus on emotion and human connection through a broader artist story will remain a powerful way to connect and create communities.
This is what I am most excited about. Once all the hype and fear fades a bit, I think the new era will show a lot of promise for true artists and the communities that will be created by authenticity.
Also, as a side note, I think that the experience of live music with humans on stage and in the crowd moving together in harmony has the least to fear regarding disruption. To me, it is one of the greatest feelings in life.
We Need Authentic Human Artists
The fact is we need authentic artists as much as ever. With all the conflict and turmoil and imbalance in the world, art is a voice and a unifying power to pull us together and put us on the right direction.
Technology will distract, but also empower.
In the end, our society is made up of humans and together we get to collectively choose our future.
I’m excited and inspired by all the creativity and innovation that I see online daily in our new world of interconnectivity.
With compassion and creativity it will only get better.
~ Michael
Exciting Tech of The Week
I already talked at length about Udio and Suno AI, the text to music generation apps that have exploded onto the market in recent months. You can play around with them and see for yourself at the links below:
Udio
https://www.udio.com/
Suno AI
https://suno.com/
My Creative Updates
I previously mentioned that I was working on a new song using an upcoming show date as a deadline. Well, the show got pushed to May 4th (giving me a little extra time) and I finished it!
I know, people finish songs every day, but my constant battle with perfectionism makes it a reason for excitement and celebration for me, haha. I’ve been keeping with the theme of staying simple with just acoustic guitar and vocals and its been a great exercise. I love production and synths and all that good stuff, but I’ll make my way back to that after getting a little better at conquering the basics.
On the more technical creative side, I’ve also been working on an exciting new feature for Niche Creator which I’ll be talking about more soon.
Lastly, it’s been a ton of work since the beginning of the year and not much time off, so I’m excited to be taking a little time to get away this weekend to Joshua Tree just outside of LA. It’s literally only a weekend, but I’ll make my way up to a proper vacation one day. I’m always so energized and inspired by exploring nature even if only briefly, my goal in the future is to make sure that I do at least a long weekend every 4-6 weeks. I’ll get there…