Paris-headquartered Believe says it’s working to keep music that’s fully AI-generated off its platforms – and the technology to detect it almost perfectly already exists.
“We have deployed a number of quality controls in our business and we aim not to distribute any content that is 100% created by AI, whether that’s through Believe or through TuneCore,” Believe co-founder and CEO Denis Ladegaillerie told analysts on the firm’s Q1 earnings call last Thursday (April 27).
TuneCore is Believe’s DIY distribution and services platform that enables artists and labels to upload their music to digital services providers (DSPs) like Spotify, Amazon Music, Tidal, Deezer and others.
The arrival of AI-generated music has triggered a number of concerns for the music industry.
One of those concerns: the amount of music being uploaded to DSPs is growing rapidly, and music companies – as well as DSPs themselves – have expressed worry that AI-generated tracks could soon flood the market and dilute the prominence and earnings of human artists.
Last week, Sir Lucian Grainge, CEO and Chairman of Universal Music Group, warned that “the recent explosive development in generative AI will, if left unchecked, both increase the flood of unwanted content hosted on [music streaming] platforms and create rights issues with respect to existing copyright law in the US and other countries”.
On Believe’s earnings call, Denis Ladegaillerie was asked by Goldman Sachs analyst, Lisa Yang, to detail the controls Believe/TuneCore has in place to detect AI-made content, and the accuracy of these processes.
Ladegaillerie responded that the technology to detect AI tracks already exists, and that Believe is working with AI companies to deploy it on its platforms – something he says is just one or two quarters away from happening.
“You have technologies out there in the market today that can detect an AI-generated track with 99.9% accuracy, versus a human-created track,” Ladegaillerie said.
“Something … we feel very good [about] is the fact that the ability to control [AI uploads] is there. Now it needs to be deployed everywhere.
“We need to finalize the testing, we need to deploy, but these technologies exist. How much time is it going to take to put into full deployment? Probably a quarter, maybe two.”
Detection and ‘blocking’ of content doesn’t cover the full extent of Believe’s foray into AI tech, however.
On the Q1 earnings call, Ladegaillerie said his company is now looking for new ways to monetize music that users of generative AI create.
Yet what happens if a new track has been created via AI platforms that have been ‘fed’ existing copyright recordings (for example, the voice of a star artist, duplicated via the computation of their past performances)?
Should a certain percentage of royalties from this “derivative” recording be paid to its rightsholders for plays of the new track?
Believe thinks so, and says it’s already “experimenting” with tech that can determine the “attribution” that should be claimed by rightsholders of a recording that then directly informs a new AI track.
“We view AI as an opportunity… to augment human creativity, it’s also an opportunity for monetization for the artists, especially with generative AI,” Ladegaillerie said.
“So what we are doing now, we are experimenting with a few of the large global AI companies around attribution… If a track is being created tomorrow that uses music from several artists, if we want to be able to license that track [to someone using it to create a new recording], for the original artist to be compensated, we need to know what percentage of the new song that has been created is attributable to this artist or that artist.
“And so we are doing a number of experiments to see how reliable attribution models are, so that we can move them potentially to licensing and generate a revenue opportunity for the artist.”
“AI technology is here, in our view, to enhance human creativity, not to replace it.”
Denis Ladegaillerie, Believe
Ladegaillerie added: “We’re also experimenting around how AI can be used to deepen the engagement between audiences and artists.
“We know that there are a number of users that like to create content [based around their favorite] artist. That’s their way of engaging with the artist.
“So we’re doing some tests in that area as well with some of the larger [AI] players. At the end of the day, we think this is going to create an additional revenue opportunity for the music industry globally.”
Later in the call, Ladegaillerie compared the tech required for successful AI music “attribution” technology to YouTube‘s Content ID, which automatically detects copyrighted elements within videos uploaded to the service, and then either blocks the video, or monetizes it with ads, depending on the copyright owners’ Content ID settings.
“It’s really about analyzing a track, analyzing a voice, analyzing how a new track has been made up, what various components were made up from other tracks,” Ladegaillerie said.
“And we do think that there’s a couple of attribution models out there that are sophisticated enough, or not far from, to be able to really have very precise attribution models [in AI-generated tracks].
“So we do think that, yes, we will be able to say this track is X percent this artist, X percent this artist, X percent this artist. We do think that the technology is actually here today, it’s just not yet been deployed.”
Earlier in the call, Ladegaillerie expressed Believe’s general philosophy on the issue: “AI technology is here, in our view, to enhance human creativity, not to replace it.”
Believe’s earnings call followed the release of upbeat numbers for Q1 2023. The firm’s total revenue hit €198.6 million (USD $213.09 million) for the quarter ending March 31, a 22.2% YoY increase.
The company’s “premium solutions” division, which includes its core premium label and artist services operation, saw revenues of €186 million ($199.57m) in the quarter, a 23.0% YoY increase.
Its “automated solutions” division, which includes TuneCore, saw revenue grow 11.2% YoY, to €12.7 million ($13.62m).
That includes a 40.0% YoY increase in revenues in the Asia-Pacific/Africa region, and a 25.2% YoY increase in the Americas.
Believe is one of a growing number of businesses centered around music that are working with AI technology to enhance their business.
In February, Amazon Music announced a “collaboration” with generative AI music company Endel to create wellness playlists. Endel runs an ecosystem of AI-powered apps that produce personalized sounds to help people focus, relax and sleep.
In April, Deezer CEO Jeronimo Folgueira said during a company earnings call that Deezer is using AI for its own wellness app.
In March, Warner Music Group participated in a funding round for AI music startup LifeScore, which creates adaptive, on-demand music “algorithmically tailored to the context and needs of the listener.”
Concerns over the potentially negative impact of generative AI on IP-based industries led to the creation of a new industry group earlier this year.
The Human Artistry Campaign – which includes the Recording Industry Association of America (RIAA) among its founding members – aims for the responsible use of artificial intelligence “to support human creativity and accomplishment with respect to the inimitable value of human artistry and expression.”Music Business Worldwide