AI-generated music and other creative content continues to grab the headlines across the world, from the fake Drake scandal to the current SAG-AFTRA / WGA strike that is threatening to grind Hollywood to a halt. The rush to commercialize the facility of AI-generated work is proceeding at a pace where current laws can’t keep up.
The rub, of course, is that AI can’t generate anything without a body of existing work to sample from. And, that work needs to come from actual human beings. A recent study, for example, concluded that AI-generators that are fed a diet of only other AI-generated content will eventually contract something researchers at Rice and Stanford University are calling “Model Autophagy Disorder” or MAD, and the results continue to deteriorate over time — just like the inbreeding Hapsburgs.
Keeping Composers In The Picture
Hollywood’s unions may have some clout. But, who is keeping the fate of human music composers in mind?
The issue of how the original creators fit into the picture is one that Alex Mubert, Co-CEO and founder of Mubert, has considered from the beginning. There’s a good reason for that. “Both music and IT,” Alex expands on his background. “I have a jazz music education.” His co-CEO, Paul Zgordan, is also a DJ and musician.
Alex Mubert, aka Alex Kochetkov, has an educational background in both computer science and music, with a Bachelor’s degree in Mathematics and Computer Science, along with studies in computer software engineering, and a degree in Business Administration and Management. He studied bass and contrabass at Jazz College in Moscow.
After his years of study, he took on a variety of roles in different companies with a tech focus. “I have worked as a freelancer and creative director,” he says. That includes working with large international clients like Nike.
But, even as he pursued his IT goals, he kept his heart in music. “All this time, I was playing music.”
Naturally, he also began to experiment with ways of bringing his two passions together. “I was trying new ways to compose music,” he says. In particular, he wanted to be able to generate music non-stop in real time. It’s a concept he calls functional music — the background to an event of really any kind. In addition, it can operate at the level of the individual listener — i.e. each listener will hear their own personalized non-stop feed.
After creating the algorithm together with a collaborator, he took his show on the road, so to speak, to the US to get venture capital.
Launched in 2017 with one of the first publicly available music generating AI applications, Mubert first established relationships with music creators and composers, who are paid for their contributions to its sampling library. Mubert has a current library of about 2.5 million sounds, including loops and full tracks, it has legally purchased for use. That’s the database that its AI has to work with.
His vision is of an app that will allow people to easily experiment and create music on their own. Because of licensing complications — i.e. AI-generated music does not fall under conventional copyright rules — the music people create using Mubert cannot be released on Spotify or other streamers. He notes that the licensing for classical music, in particular, complicates the issue. It can, however, be used in videos, as background to podcasts or interviews, in advertising, and other commercial applications.
For listeners, there’s another perk. “You can listen to this never ending radio.”
“A year ago, we started a new application for content creators,” he adds. The company also launched one of the first text to music applications in late 2022. The user-friendly moves have led to the company’s rapid growth. “We’d like to continue to focus on creators,” he says. As he points out, the text to music application allows people who don’t have a background in music theory to generate music.
The company offers a suite of products designed for different users, who can generate music to suit their needs, from 15 seconds to 15 hours. Mubert API is a paid platform for B2B clients, Mubert Play is focused on listeners, and Mubert Render was launched in 2022 for non-commercial users. Based on anything from an abstract phrase to a mood or genre, Mubert Render can generate a track up to 25 minutes in length. According to company statements, Render has more than 100,000 monthly users who can generate tracks in more than 150 different categories based on mood, genre, or activity.
Not surprisingly, because of the current situation, the majority of Mubert tracks fall into the Lo-Fi, Ambient, and Chill genres, which are suitable for use as background music for many types of applications, such as podcasts, livestreams, interviews, among others. Classical music, he says, is another one of the popular genres used by content creators to match music with their video and other multimedia works.
The numbers, according to a company statement, are impressive. Mubert claims its users have created more than 56 million tracks, and started 44.2 million streaming sessions. The Mubert catalogue now contains more than 100 million songs that have been created using only its licensed material as input.
About The Musicians…
While the focus in terms of users is on those who will use Mubert to generate and listen to music, Alex says he hasn’t forgotten about the original creators. “In my mind, human music will always be more amazing than AI-generated music.”
For composers and music producers, in addition to the bank of music the company has purchased outright, he’s currently developing a royalty-based model that he hopes to launch soon. Instead of the company buying tracks and loops outright, composers/producers will be able to upload their own work, and then receive royalty payments each time it’s used in generating a new piece by an end user.
It’s a field where the landscape changes seemingly on a daily basis, and staying on top of it will prove key to managing the future. “When I started this start up, when I mentioned AI-generated music, the investors were like, what is that?” laughs Alex Mubert.
“Me as a musician, we always look for solutions that support musicians,” he said. He’s looking, in other words, to create a 21st-century business model that will support all the stakeholders. “It’s one of the goals — to create the new business model for both.”
His prediction for the near future? That Spotify and other streamers will reconsider their current ban on any AI-generated music. “They do not want to do it now, but they should think about it,” he says.
On top of that, he points out, the mainstream end of the music biz still largely ignores the rapidly growing so-called sync music industry, where music is licensed for use in movies and videos. That’s a huge area of growth for a service like Mubert.
Looking beyond that issue, the future, as Alex sees it, will lie in real-time content generation, the “AR glasses, games, Metaverse stuff” as he calls it. As people enter their own virtual realities, they’ll be accompanied by a personalized soundtrack of music that will be generated in real time. That’s a reality where the old business model falls away all together.
“It cannot be solved with playlists and Spotify. It can only be solved with AI.”
#LUDWIGVAN
Get the daily arts news straight to your inbox.
Sign up for the Ludwig van Daily — classical music and opera in five minutes or less HERE.
