top of page

How Licensed AI Can Grow Music Revenue (Live from Music Tectonics 2026)

Writer: Evan Nickels
Evan Nickels
11 minutes ago
33 min read

What happens to the value of music when AI is licensed, consented, and paid for? 

 

This week, we’re sharing a panel from last month’s Music Tectonics Conference titled, “Music is Worth More: Tech that Multiplies Revenue, Not Disrupts It.” Moderated by Dmitri, this panel includes

  • Ana Guillén, SVP of Go-to-market at Stability AI, whose Stable Audio models are trained entirely on licensed music

  • Bridget Ferris, who leads Talent Partnerships at ElevenLabs, a voice AI company with a marketplace that pays contributors every time their voice or music is used

  • Anne Kavanagh, CEO and co-founder of Royall, a platform helping artists and creators manage and license their name, image, voice, and likeness

 

They get into where new money actually comes from, including a voice marketplace that paid out more than $11 million in 2025. They also cover whether artists get paid when models learn from their music, who owns a track made with AI, and what labels, publishers, and managers can do next. 






Listen wherever you pod your casts:



Looking for Rock Paper Scanner, the newsletter of music tech news curated by the Rock Paper Scissors PR team? Subscribe here to get it in your inbox every Friday!



Episode Transcript

Machine transcribed


[00:00:00] Jade: And now I'd like to introduce Dmitri Vietze, CEO of Rock Paper Scissors and Music Tectonics founder, who will be moderating this session, Music Is Worth More: Tech That Multiplies Revenue, Not Disrupts It

[00:00:18] Dmitri: Thank you. Thank you all. Thank you all so much for coming Yep, I'm Dmitri. as most of you know, we have a theme to the conference this year, Rising Tide, Grow the Music, Grow the Value. And I specifically came up with the topic for this session early on in the programming because I really wanted to focus on what future technology is doing that.

I looked at eight different revenue multipliers of technology that I felt like would increase the value, increase the TAM. Uh, and I specifically wanted to find licensed AI companies because I feel like there's so much growth potential in a new way. We know that streaming is kind of plateauing, and there's opportunities for what else could happen with premium subscriptions for streaming, and there's a lot of other additive things that are coming into play as well.

But I actually pulled licensed AI separate from other types of licenses. I think there's other growth opportunities for licensing and sync, new types of ways that music are being used. We've seen it in fitness and wellness and gaming. There's all these cool things that are emerging. Music goes with everything.

But I felt like the way that AI works is different than all that other licensing 'cause something transformative is happening there, something maybe derivative depending on how it's being used and so forth. And so we really wanted to look into that as its own category. We're like, "What if we get this right?

What if we get licensing, attribution, monetization with AI right? How does that grow the value for the music industry overall, for artists, labels, publishers, the whole teams and so forth?" And then I always look at the, Pophouse model of ABBA Voyage and think about the fact that a, company that normally would buy publishing rights and then monetize songs is now creating new performances, and they bought the Kiss catalog as well.

And they... I don't know if you know this, but when they bought the Kiss catalog, they didn't just buy the name, not just the, the rights to the songs, not just the name, image, and likeness of Kiss. They bought the biometric data of Kiss as well, which means they can create new moves that the original artist never did before, never made.

They have the song rights. There could be endless Kiss songs that they now have the ri- I don't know if that's what it looks like, but potentially somebody could buy the rights to the point where they could have endless songs of a band that people already love, that already has a following, that already has a look.

I'm imagining the, future Kiss shows at Pophouse. People are probably gonna either come in face paint or they're gonna pay to have their face painted. You can sell merch or you could get your face painted. You can have an whole experience. When I was a kid, we would go to the Rocky Horror Picture Show, and everybody knew all the words.

They showed up in full dress. they threw rice during the wedding in the movie theater. And I'm imagining what are the experiences that you could do once you unlock this new type of licensing? So we have people here on this panel that represent that world of licensed AI, where there's...

Licensed means that they're paying to use it, right? And they're working with labels in such a way that the labels are saying, "Oh yeah, there's a possibility down the road." There's opt-in agreements with artists. There's all sorts of things, and also somebody here who is helping us figure out how do you manage the whole name, image, likeness, biometric data component as well.

Little more intro here. I'm getting to it, guys. I'm getting to it. Goldman Sachs just published its 10-year Music In The Air report. It projects the global music business growing from $112.5 billion to nearly $200 billion by 2035. Their report includes live, so almost half of that is live-related stuff, whereas other reports it's, a lot of times it's like recording and publishing and so forth And it says the industry's focus is shifting from consumption growth to monetization growth, meaning not getting more listens, but how do you make more money off of the listens or whatever's happening with music out there?

 Its examples are new subscription tiers and add-ons such as AI-powered remix tools and direct-to-fan sales it calls additive to streaming, not cannibalistic. There's stuff in there. But on AI, Goldman sizes a base case scenario of about $3.3 billion of new revenue by 2035. But I, I think that number was built on today's paid streaming subscribers.

I don't think it fully accounts for the potential for licensed AI platforms like the ones on this stage today. I think that's like, "Oh, let's pay extra so we can remix on Spotify." That's not who's up here. That's not what... This is, this is new stuff. this is really interesting. So it's about, this panel's about the idea that music is worth more than its current market size, and that tech doesn't have to be disruptive to the value of music.

Even in the Wild West moment that's AI, let's get into this. Let's get into how we could get it right and what the value of music could grow using this. Ana.

[00:05:06] Ana: Yes.

[00:05:07] Dmitri: With Stability AI, she spent her career where creativity meets technology. She spent eight years at Warner Music Group, managed artists and digital creators at a large management company, and led creator marketing at Instagram.

Now she's SVP of Go-to-Market at Stability AI, whose stable audio models, which are super fun to try by the pool, are trained entirely on licensed music, yay, and which is backed by all three major labels. Facts about Ana: she's currently 66% of the way toward her personal goal of running all six world major marathons.

[00:05:41] Ana: Yes.

[00:05:42] Dmitri: Crazy. And she once opened for Morning Jacket, and her Spotify listening age is 17. Okay, Taylor.

[00:05:52] Ana: I'm now embarrassed by these fun facts. I was trying to be unexpected, so I really went out there.

[00:05:56] Dmitri: What's up with the Morning, My Morning Jacket thing?

[00:05:58] Ana: I was hoping you weren't gonna ask me. I, when I was working at Warner, I was working with a band, and they were opening for My Morning Jacket, and they had a song where they needed a whistler.

They had someone that was gonna whistle-

[00:06:10] Dmitri: That is sick

[00:06:10] Ana: ... and that person didn't show up. I just so happen to be a really good whistler. And so they asked me if I would go on stage, and so I technically feel like I opened for My Morning Jacket. You guys let me know if that is true, but-

[00:06:23] Dmitri: I think you did ...

[00:06:23] Ana: that's the story.

[00:06:24] Dmitri: I kinda wAna see how your whistling is.

[00:06:25] Ana: I knew you were gonna ask me that. I, I can't do it on the spot.

[00:06:28] Dmitri: Okay.

[00:06:28] Ana: The performance pressure's too much. In the wind. Give me, like, a couple drinks after this-

[00:06:31] Dmitri: Okay

[00:06:31] Ana: ... and then I'll get on the mic.

[00:06:32] Dmitri: At, at the pool party, we're gonna have her whistle.

[00:06:34] Ana: Perfect.

[00:06:35] Dmitri: So Ana, Stability decided to train only on licensed music and to build for professionals rather than consumers.

Why did that become so important, and how did you land there as a company?

[00:06:45] Ana: Yeah. Stability AI is best known for being the developer and releasing Stable Diffusion, which is maybe still one of the most popular text-to-image generation models. We've since evolved from image, and new leadership came on two years ago, and we're led now by our CEO, Prem Akkaraju, and our chairman of the board is Sean Parker, who is best known for Napster and Spotify and Facebook.

Prior to Stability AI, they led Weta Digital, and Weta is a large visual effects company that works on some of the films that we all know and love. They've built software for professionals at the highest level. And so when they came to Stability, they reoriented us and they focused on professionals. And there's really two verticals that they focused on.

One is interactive. We're partnered with Electronic Arts, which is a large AAA game developer, and we're helping them sort of rethink game development using AI in, ways that really empowers their artists and their designers. And then the second is audio. And in the audio space, obviously Stable Audio, you should check it out over there at the pool, backed by the major labels, Sony, UMG, and Warner, and we're also developing technology with them.

 the reason why we're focused on professionals and why they shifted us this way is because their belief and my belief is that if you can develop tools and software that a professional creator wants to make, you can develop tools for anybody because a professional is very discerning, very critical in the best way possible.

And so we wanna be at, like, the tip of the iceberg, develop tools for the best of the best, and then we'll see where things go. But today, that's where we're focused.

[00:08:21] Dmitri: Got it. Great. Thank you. I like to do this thing where we introduce people as we go. So Bridget Ferris has been on the artist side of the business her whole career.

She was a talent agent at UTA, then spent 14 years at Universal Music Group. We've got these label veterans here, where she was VP of brand marketing, running campaigns across its artists and labels, and she joined ElevenLabs this year as part of a group of music industry veterans brought in to shape how the company works with artists, and she leads talent partnerships for ElevenLabs and Eleven Music.

Fun fact to know about Bridget, she's a twin. I'm also a twin. Her brother also works in tech and is frequently heard as a news broadcaster on San Francisco radio stations. That's pretty cool that you're doing this voice stuff and he's doing voice stuff, too.

[00:09:06] Bridget: Yeah, and I just realized my fun fact is about him, so I probably should think about my own

fun fact.

[00:09:09] Dmitri: Twins, twins be like that sometimes.

[00:09:11] Bridget: It happens. Yeah.

[00:09:11] Dmitri: That's just how, how it goes. You were in the womb together, so.

[00:09:14] Bridget: Yeah. I can ta- I can get credit for his broadcasting career.

[00:09:16] Dmitri: Since I've got a, uh, twin sister, do you get that question, "Are you identical?" all the time?

[00:09:20] Bridget: It's like- Way too many adults ask me that who should know better.

[00:09:24] Dmitri: Fraternal twins can't be identical.

[00:09:26] Bridget: Yeah.

[00:09:26] Dmitri: Think about it.

[00:09:26] Bridget: It's your biology fact of the day.

[00:09:29] Dmitri: So you came up on the artist and label side. How does that background change the way you approach AI compared with someone who came straight from tech?

[00:09:37] Bridget: Yeah, I think, a normal question that if you're on the tech side and you're leading with marketing, you're gonna think, "What can this artist or this talent or celebrity persona do for our tech and our brand?"

 but the way that we approach it at Eleven Labs, and the reason why I joined the company, is they really wanted to see what can the tech do for the artist and, and for these teams and these creators. So that's why w- they brought in a lot of music industry veterans, and folks across the landscapes, both at the intersection of, tech and music as well as just traditional music, just to make sure that we were providing that point of view and that infrastructure across how our deals and how our models are created, so that way it makes the most sense for artists and the many, many companies and labels and management, groups that service them.

[00:10:21] Dmitri: Nice. Great. So we've got our two licensed, music companies here, and also you get into some of the name, image, and likeness stuff because of voice as well. Anne Kavanagh here is a 20-year entrepreneur from Ireland, now based in New York. She founded a music streaming app for rideshare passengers that now works with the major labels on new releases.

Pretty cool new use of music. Then helped build podcast ad tech that Spotify used to monetize its long tail. Now she's CEO and co-founder of Royall, with two Ls, which lets anyone from a superstar to an independent artist register, protect, and license their face, voice, and motion. Fun fact, one of her co-founders was her first boss.

You had to do it. And she met the other online about 15 years ago through a podcast project, and they've built multiple businesses together since. That's pretty cool. You must have liked your boss and-

[00:11:13] Anne: Well, I hope so at this stage, but also similarly, I'm like, "How do we all follow up after Ana's fun fact?"

 

[00:11:20] Dmitri: Yeah. You could whistle.

[00:11:22] Anne: I can't.

[00:11:24] Dmitri: What did you see coming that made you build Royall?

[00:11:27] Anne: Yeah, I mean, I've spent pretty much the last 10 years in, you know, working with the long tail or emerging creators, both on the artist and spoken word side, and have been in this position where, you know, we've experienced content creation, distribution, and monetization.

And you know, with our music company, we've literally listened... Somebody on our team listens to every single song that comes in. And last year we started to think about like, "Well, what happens when the human ear or eye can't tell the difference between real and synthetic content? How do artists, you know, protect themselves, but also their income, right?

Or creators at a whole?" And so, you know, we just started to think about like how do we build a solution around that within our own ecosystem, and then as we started working on it, we were like, "Oh, there's a much bigger opportunity here," which is every single one of us that are sitting here right now are either at risk of our likeness being used on the internet, or if you're commercially inclined, there's an opportunity to participate in it.

And I think the way that we think about, how content creation and distribution and ownership of likeness- is working is similar to, you know, in, in music for example, has, infrastructure that already exists. You distribute your, you know, your track, it's registered, there's tracking of usage, it collects the royalties, and we believe that that should also exist for, likeness because creators now become the content instead of create the content, right?

So it's more fluid as to what your participation is in media or content at a whole as we move forward.

[00:13:10] Dmitri: It's like we could all become the ABBA Voyage, right? And also, it's interesting to see what's happened with name, image, and likeness on the sports side, and now it's starting to translate even... Well, we need some infrastructure for it, clearly.

But on the music side as well. So three companies that want the artist paid. So let's imagine this works. Let's imagine that we figure it out correctly and, and this is the world of music that we're moving into. Let's say we get it right. What does it look like for a working artist, say, three years from now?

Ana, you wanna start us off?

[00:13:43] Ana: Sure. Working artist three years from now. I think there's a couple things that I think about when it comes to generative audio and, and being an artist. One less associated to revenue, but I think it's really important and we talk about this a lot. We talk about the notion of moving from prompting gen audio to playing gen audio.

I think Prem, our CEO, spoke about this a little bit yesterday, but, like, what musician do you know when they wanna start creating goes and, like, writes down exactly what they wanna create? Because, like, how do you describe music with natural language? It's a feeling and it's an emotion and, the industry needs to move in the way that models are trained and the way that products are built towards the way that musicians actually create, and that means, like, playing.

So you're not starting with natural language, you're starting with your voice, or you're starting with MIDI. And how do you think about interacting with the technology in, like, a tactile way? And so that's one thing that I think is definitely happening in three years from now. I think the way that we interact with generative audio will be completely different than what we know today.

The second piece of that and sort of building on top of it is we believe that artists will have their own models, so they'll have a palette of models that are trained on their sound, and that's what artists will be using and that's a way for them to scale their sound. It's a way for them to monetize.

I'm actually curious for the audience, when I say open-weight models, does anybody know what that means? A few people. I'll explain it just a little bit. So at, at Stability AI, Stable Audio is an open weight family of models. And so there's four models, three are open weight, one is closed weight. And an open weight model means that you can go to Hugging Face and you can download the model.

It's a big model repository, sort of like a GitHub for models. And you can download the model, and what you're able to see is, like, the ingredients. You're able to see the recipe of how the model was trained. You can build on top of that if you're a developer, or if you're an artist and you have your own sound palette, your own catalog, you can actually take that source material, bring it into the model, retrain the model, and then the model can then generate in your style or in your sound or whatever you want it to sound like.

And we believe that's gonna be a way for artists to monetize moving forward.

[00:15:58] Dmitri: I love that. So their whole style of music becomes a new source of creativity, a new source of interactivity with others, and a new source of revenue at some point as well.

[00:16:08] Ana: Yeah, exactly. And one thing I'll, add on that is, like, when you think about the, ethos of music, it's collaboration, it's remixing, it's interpolation.

Like, there's always this sort of, like, blending of sounds and styles and artists. We think that the ethos of that is very similar to fine-tuning a model, and sort of like merging two things together and then being able to create something net new as a result of that.

[00:16:30] Dmitri: I love hearing you talk about it, 'cause as we get to ElevenLabs and talk about each one's a different lens.

You guys do things differently, your base is a little bit different, and there's a different answer to the question. So, Bridget, I'd like to ask you the same thing. What does it look like for a working artist three years from now looking through the ElevenLabs lens?

[00:16:47] Bridget: Yeah, I think for me, it's all about creating more options for the artist.

So if you are a frontline artist, um, I think every... and maybe even sooner than three years, having interactivity and AI, functionality will be a, critical part of your marketing rollout strategy. and being able to, like, work with fans and getting a deeper consumption with the music, is gonna be a, really important way that the next generation and the next evolution of music consumption works.

 whether that's everything from, like, you know, remix competitions, all the way to allowing users to reimagine and maybe even, collaborate with the artist directly, through this technology is gonna be the way things are gonna move. And, uh, what's nice about that is that the artists are not just, you know, in the background or just providing the, frontline track, but actually able to, you know, monetize all those derivatives and be able to actually participate in that creation process, so they're not just the source of the training material, but they're also benefiting from it and able to use and potentially monetize the output that fans work on with them.

 and then if you are not a, frontline artist, but you're just a working musician, and you have, , really amazing stuff and you wAna be able to monetize, I think there's gonna be way more direct to consumer, direct to businesses, direct to creator places. Like, we have a marketplace where musicians can post their music, and, brands, marketing teams, agencies, individual creators can choose to directly license that, track based on whatever needs that they have, whether it's social, paid, broadcast, et cetera.

And that revenue goes directly to that music creator. So, that we cut out the middleman of, like, a big, you know, sync arm, or there's , this need to have this muscle of this representation. Like, you can actually create that and set yourself up to create a lot more passive revenue.

[00:18:28] Dmitri: I'm curious about fans' interactivity as well.

Like, for artists to see their fans engaging as well, I'm, I'm curious if you see a future there, as it relates to what ElevenLabs is up to?

[00:18:40] Bridget: Yeah, like fans interacting with, music.

[00:18:41] Dmitri: Yeah.

[00:18:42] Bridget: Yeah, I think, um, I think there's definitely gonna be places where music is gonna be, again, like a, something that people are gonna they're not gonna passively just put on a track and listen to it.

There's gonna be places and modes for, uh, and again, set with the artist's guardrails and artistic integrity in place, allow fans to reimagine, or mash up their own versions of the tracks to better serve their own personal listening. Whether they wanna slow down lo-fi track for studying, or if they want something that's, like a disco version of their favorite ballad.

 there's gonna be a, a lot of different ways that fans can then interact with the music, more than just the track or the remixes that are put out by the label.

[00:19:16] Dmitri: Yeah. Nice. Thank you. Anne, let's ask you as well, what does it look like for the working artist if we get it right through the lens of what Royal's doing with, this infrastructure?

[00:19:25] Anne: Yeah. I mean, to follow up on what Bridget said is, like, we've got very similar mindset or philosophy around, the creator being able to participate in other user experiences. But actually to speak to the fandom piece is, you know, how do you... If you use your favorite artist in a remix and you generate a cute music video with you in it, and then you distribute it on the social media platforms, but maybe it's two of your favorite artists, how do you- Track the usage and the monetization of using that likeness, whether it's voice or face.

But also as well as, like, you know, licensing. So I think there's an opportunity there and, you know, our infrastructure allows any artist, big or small, to be able to set their consents, their permissions, and also create licensing around how it can be used, whether that's, you know, right now it's external of platforms, but, you know, conceptually is where it can be integrated into any platform that you're creating from the moment of creation, right?

So I can pick Dmitri-

[00:20:31] Dmitri: Mm

[00:20:31] Anne: ... and say, "Hey, I'm going to create an experience for myself or for somebody else. How do you get, financially incentivized and give permission?" And I think even on the fandom piece is like, you know, maybe there are situations too where artists are like, "Okay, cool. I don't wanna monetize from this, but I wanna use it as a marketing opportunity, and I wanna bring my fans into that experience."

You still need to set controls and guardrails around how that can be used, right? and I think the other thing is like, in terms of the training models and licensing of like catalog. You know, you think about, funds or buying catalog of music right now, you're buying the catalog.

Does that mean you own the voice? So how does that translate when we start thinking about, you know, buying catalogs or even just again as like the asset being used in other ways outside of like traditional formats currently. so yeah, and I think to go back to it is like some artists want to participate, which is fine.

They should be able to monitor and, facilitate, like, takedowns and all the rest of that. But for those who do wAna participate commercially in it, I think there's a massive opportunity there.

[00:21:43] Dmitri: It's so wild to hear you talk about all these different... It's almost like different types of rights and use cases, and the parallels that we saw in other parts of music, it's a lot more complex-

[00:21:54] Anne: Yeah

[00:21:54] Dmitri: because there's so many different possibilities. There's so many contingencies of what could be happening, with these new-

[00:21:59] Anne: Yeah

[00:21:59] Dmitri: ... new use cases. And here we are talking about kind of like the training of AI for the generative musical uses, but there's probably a lot of opportunities for AI to actually make the infrastructure work much more flexibly.

We had a session earlier on MCPs and, and h- you know-

[00:22:15] Anne: Yeah

[00:22:15] Dmitri: ... on the infrastructure side for labels and stuff, but I think there's also on the infrastructure side for any-

[00:22:20] Anne: Yeah

[00:22:20] Dmitri: ... GenAI companies or, or licensing-

[00:22:22] Anne: Yeah

[00:22:22] Dmitri: ... uh, infrastructure too.

[00:22:24] Anne: And I think, just one last thing is, like, diversifying outside of, you know, for artists, their monetization comes through, royalties, merch, ticket sales.

Now, there is an extent, you know, another opportunity on NIL, which is your favorite artist's voice can now be used, GenAI, in a radio ad or... So they can now multiply the... They can scale their talent in ways that they haven't been able to before and create other opportunities because... And also the extension of, like, audience size, right?

So their actual likeness is the asset versus the audience size that they're reaching.

[00:23:01] Dmitri: I just remembered that thing where artists would go on tour and they'd have to do, like, 50 radio spots for all the tour cities, but this is even... I mean, sure, they could replace that with the NIL-

[00:23:13] Anne: Yeah

[00:23:13] Dmitri: ... rights. but they could also be doing custom commercials for something totally unrelated to them without actually have to go into the studio, which was, is very scalable as well.

Well, everyone who's been following my sessions in the Rising Tide podcast series that I do knows that I've been collecting billions of dollars. I've been collecting billions from background music, from gaming, from, you know, all the new licensing and AI licensing is one of them, where I'm trying to...

That's where I'm pushing things. So I wanna ask the money question. Where does the new money actually come from? When we look at this, getting this right, I'm curious from each of you, like, you know, there's gotta be a source for the money, right? Where is it? And Bridget, let's start with you 'cause you guys have already done some interesting things , on the voice side as well, so you have kinda like this track record, as well.

But where does the money come from?

[00:23:58] Bridget: Yeah, we have a m- I'm gonna use the word model, but I don't mean it in the LLM sense. I mean like we have a, a-

[00:24:02] Dmitri: A business model? ...

[00:24:03] Bridget: a business model that it's proven. so our company, Eleven Labs, started out as primarily a, dubbing company.

 we still do a lot of work in that space. It's something we're really proud of. part of that is our voice library where we have a collection of voices that, our different users can tap into for their projects. , Some of those are fully synthetic. Some of them are actual humans' cloned voices who've given full consent for that.

Um, and the way that we wanna make sure we are compensating these folks who lend their voices and post them to our marketplace is to develop this system. So we have our voice library, um, where if anyone today could go on and create a professional voice clone, list it, on our voice marketplace and on, on, within our voice library, and if a creator or a brand or a marketing team likes your voice, um, maybe you speak a particular language that we don't have a lot of, or you have a really particular, you know, DelCo Philadelphia accent that's really, beautiful for someone to use.

 I can say that 'cause my husband's from Philly, and it's a cross I bear. But, um, every time that that voice is used, that, voice actor, that voice creator is, compensated for its use. Um, just last year alone in 2025, we paid out over $11 million to, individuals who lent their voice.

I think we're pacing at over $22 million total currently. Um, and so that's that same model that we're replicating for the music space. So I mentioned on my last answer this music marketplace that we've developed, and we launched it earlier this year, where creators who create tracks can make them available, so that way w- , marketers, our enterprise customers, our individual contributors, if they have a piece of maybe it's, visual first media and they just need a song that's commercially safe, rather than going on and creating one, which they're welcome to do, but a lot of people don't have either the time or the artistic ability to capture what they want, but these amazing music creators do.

So they can then search into our marketplace for a song that matches the vibe that they want, and depending on the terms that they want, they can access that song. And again, those dollars go straight forward to the, uh, creator in question. And it's a... It sets up like a passive marketplace, so it's kind of like how you think about Etsy.

If you put a, document or a wedding invite template, you just, you can just make that money as you sleep. and it's totally up to you what the terms are. You can take it down at any time. but you kinda commit upfront for a certain period, and then it's, very much compensation and consent forward.

[00:26:26] Dmitri: $11 million is real money. That's new money coming in, you know? If we do that in music too, we gotta get to a billion before the end of this panel, guys, so we got a lot of work to do. But honestly, I feel like it's really early for any kind of licensed AI-related revenue, but it gives me hope 'cause that's a number that make, that makes your head turn.

It's like, okay, there's new money that those people couldn't have, collected elsewhere before. Anne, do you have any specific dollar examples you can give us? You were talking a little bit about radio and things like that.

[00:26:52] Anne: Yeah.

 and actually on that, I think IAB just reported that 86% of marketers are planning on using gen AI in advertising.

So we see an increase in how advertisers and brands are using AI, and again, how do they engage with the creator and the content and the licensing of the music. So, you know, I think that's a really great, example of where the market is going even on the advertising side. But specific to us is, like, you know, we had a, licensing deal that, through our platform and, it's for, like, a voice for a podcast and, you know, it's a specific sound that's being...

that the brand is looking for, so audience size doesn't matter. And in that scenario, the creator, all they have to do is spend two hours generating their voice to then be recreated. It turns out at $500 an hour for two hours of work and very limited number of assets that are being created, so they have a reoccurring revenue opportunity in that scenario when the next season comes around, et cetera.

[00:27:57] Dmitri: So wait, you're saying they train their voice and they get paid $500 an hour after they've just spent two hours, it's kinda just f-

[00:28:03] Anne: Right.

That's it ...

[00:28:03] Dmitri: anytime anyone uses their voice again, they get 500 bucks an hour?

[00:28:07] Anne: So no, their voice is already created. So literally every time a new episode is created, the license is created around it, so they get paid per episode with never having to go into studio.

It's all prompt-based.

[00:28:17] Dmitri: Oh. I wanna sign up. That sounds like some passive revenue there.

[00:28:22] Anne: Yeah.

[00:28:22] Dmitri: Ana, we, we don't really have to add up to a billion, but I'm curious, from your perspective, you've talked about some different ways that music scales. Uh, where do you think the money comes from as we grow revenue from licensed, music, to AI?

[00:28:35] Ana: Yeah, I, I don't think we should count ourselves out. We should try for a billion. Like-

[00:28:39] Dmitri: Okay, let's go. Let's go

[00:28:40] Ana: ... aim high. Like, definitely aim high.

[00:28:41] Dmitri: No softball questions here.

[00:28:43] Ana: when I think about monetization as it relates to what we're doing at Stability, there's, immediate things today, and then there are things tomorrow that are on the horizon.

 immediately today, I talked about our open-weight models. Those are available on Hugging Face. Within our license, people can use that technology for free, they can build on top of it, and they can commercialize it up to a million dollars in revenue. And so we have 1.6 million downloads of Stable Audio off of Hugging Face today, and I know there's a lot of builders and developers here and, and that's our hope, is that people are gonna take this technology, and they're gonna build new applications and new ways of creating music, and we want them to earn money.

And so you can do that up to a million dollars, and once you've earned a million dollars, then we talk about what a commercial license might look like. And so that's one way today. looking forward, like very similar to what Bridget was saying, we really believe in the marketplace model. We believe that people are going to customize models, they're gonna make models off their own sound, and they're gonna sell them so that other people can be able to use them.

And so we're productizing that ability, and we'll be launching a marketplace as well, in which people will be able to generate revenue that way.

[00:29:51] Dmitri: What about an artist who isn't making new music anymore? Bridget, do you have any thoughts on where that could go? You know, like we have some folks who represent estates here or big song catalogs, legacy artists, et cetera.

What are the opportunities there?

[00:30:04] Bridget: Sure. Well, hypothetically, and I know this from being on the label side for many years, a lot of times if like, a brand or a creator is looking to license, a track that's from a well-known artist, they'll go to that estate or they'll go to that label and ask for a quote, and if it's a well-known song or a well-known artist, it's gonna be pretty high, right?

Because it, pulls nostalgia. It evokes a particular feeling. It has a lot of, gravitas to it that other tracks don't have, which is why it's a well-known artist and a well-known song. So inevitably, some brands and some creators obviously can't budget that within their scope, so what they might do, and I've seen this many times, is they'll kind of go around and say, "Okay, thank you, label," or, "Thank you, estate," and they'll go and get a soundalike at a stock or library music, destination, right?

Um, and so like if you, on, being on the label side, you'll be quoted, you'll quote a commercial and then s- three or six months later you'll see that, same commercial and it sounds exactly like your artist, but it's not your artist, which is a bummer, right? So you've missed out on that window of revenue because your actual asset costs too much and was too valuable.

So in order to help those estates and those legacies kind of capture that revenue that is really inspired and evoking from their IP, hypothetically what you could do is have your IP train, a fine-tune or a mini model, which is what we, we, we have fine-tunes within Eleven Labs and Eleven Music where you can...

It's a secure closed model where you can train it for yourself and then any of the outputs is a derivative and comes from that, actual sound, so then you can sell that for a lesser amount to kind of capture that gap that otherwise you were missing out on. So then you can participate in that soundalike versus having to go and hear it from someone else who just is monetizing off of your kind of celebrity and IP.

[00:31:46] Dmitri: Interesting. Okay. we're, somehow there's a billion dollars in there. I know there is. And also there's time, right? It doesn't have to be this year. It could be next year, right? So even with models trained on licensed data, this new question keeps coming up. Do the artists themselves get to opt in?

Where does that fit into this supply chain of gen- I'm asking hard questions now, of generative AI, and where do you think it should go, Anne?

[00:32:08] Anne: I mean, our whole business is built on consent, so, you know, I think, again, it just comes down to, like, how the artist wants to participate, number one, but how do we give the tools to the organizations who are working with these artists and creators?

 the way that we've approached it is, you know, for labels or music distros, is you actually can onboard your, roster or your community. They then set their consents and opt-ins so that when you're working with organizations like that, Ana and Bridget are with, you can see from your roster, you know, these are no-gos, and these ones are, and here are the permissions around maybe different types of platform or training use cases or commercialization.

So, you know, I think there's... And there's many companies, not just us, that are in the space that are building around it, and I think a lot of, like, individual, you know, companies like ElevenLabs and Stability are, have internal mechanisms as well. But it's how do we make it portable across the entire internet, right?

And how do we set a standard that says, "Okay, this is where the creator is, and they should be the first person who gets to opt in or out"?

[00:33:20] Dmitri: Ana, I got one for you. If I make a track on AI, do I own it?

[00:33:25] Ana: I can only speak for Stable Audio, but yes, you do. Anything that's generated with Stable Audio, you own it, you can commercialize it free and clear.

You can put it in a film score, you can sound effects in a game, you can take production elements, samples, loops, and put them into your track, and then release that on Spotify or YouTube or wherever, and it's yours free and clear.

[00:33:46] Dmitri: I love that. Some of the early models that got released at, uh, other companies was like, who owns this?

Like, what happens? Who's monetizing this? And whose distributor gets it pulled down or you know, whatever else 'cause there's a lot of, fingers in the pie, let's say. Bridget, I'm curious. When an AI model makes a track, does the artist whose music it learned from get paid? And how close are we to seeing that revenue start to come in?

[00:34:11] Bridget: That is a very actively crafted question currently. That's something we're really looking at and working with our partners, across the labels and the publishing groups and the artist teams directly on the best way to do that. I think, you know, there is gonna be, especially at the scale in which generative music and derivative music can be created with this technology, that we really need the, that infrastructure in place, like the work that Anne's team does.

So it's like making sure that we're tracking it correctly and figuring out, like what you said, is, you know, how much of the different blends are gonna make the right alchemy to monetize. But our, goal is to compensate and to make sure that the right, credits are given. but it is an active question.

I don't have the exact answer.

[00:34:50] Dmitri: Yeah.

[00:34:50] Bridget: But I s- some- we will soon.

[00:34:52] Dmitri: Ana, what about you?

[00:34:54] Ana: Yeah. So as it stands today, at the inference level, when somebody's actually generating something, that, to Bridget's point, hasn't been solved yet, and I know that there's a lot of companies that are working on that, which I think is incredibly exciting.

The way that we are, artists are monetizing on the input side, like when you think about AI, you have to think about the input side, which is the training data, and then the output side, which is what you create. On the input side, we work with large rights holders like the major labels and AudioSparks and other types of right holders that artists have been compensated to, you know, be in their library, and it's opt-in.

We have a value exchange with them, a monetary or currency value exchange with them, and that's how artists are being compensated through the models that we're training today.

[00:35:39] Dmitri: Well, and this is something you're trying to solve. how do you answer that question? Does the artist whose music it learned from get paid?

And how close are we to getting that done?

[00:35:48] Anne: you know, again, I think outside of model, so like let's say for example, the way that we approach it is we fingerprint and watermark any of your biometric assets, right? So let's say I take your voice off the internet, and I go and I use a tool, and I, I'm like, "Use Dmitri's voice in my song."

 In that scenario, you can, you know, our technology alerts you to say, "Hey, there's a 66 percent probability that your voice was used in this song." And then you can do the takedowns through it. So I say that to say is, like, it's possible, right? But I think it's like, how do we adapt that industry-wide and set a, baseline standard and conversation that happens around how do we implement that at scale, right?

So, in that scenario, if somebody wants to license your voice for a song, then they would do it directly through you. If it's an organization that's using our tool, they can do it at scale through their roster. or if it's independent direct business that you're doing as a creator, whether it be as an NIL or another artist who wants to use your voice in a song, they can do it that way.

[00:36:59] Dmitri: Let's do a speed round. If you're a label, publisher, manager, or estate that's here today, what's one thing you should do next week to start turning this into a revenue stream? Who wants to go first? Go for it, Ana

[00:37:13] Ana: I'll go?

[00:37:13] Dmitri: Yeah.

[00:37:14] Ana: I did this talk last week at Berkeley School of Music, and it was for the, an emerging tech lab, and it was for professors, faculty, staff.

And somebody asked me a question, like, "What do we tell our students? Like, how do we guide our students right now?" And what I said to them, I would say to really anybody, like I say it to my team and I would say it to anybody here, my belief right now is that you should get really good at something, or if you're already really good at something, learn how to build a system around it to scale that with AI.

And so if you're a musician, that's your craft. If you're a knowledge worker, like a designer, it might be design. Whatever it happens to be, get really good at something and build a system to learn how to scale it with AI.

[00:37:58] Dmitri: Great. Bridget?

[00:37:59] Bridget: Well, if you are a rights holder or, like, repertoire representative, I think it's really important just to, especially now as ever, is to really understand and catalog all the rights that you control, and identify the rights that you don't have access to, 'cause that'll be one of the main questions that companies like ours will need to know, just to ensure that we're keeping our compensation and consent model going.

 and then I think you're already doing the second thing I was gonna say, which is making sure you're educating yourself. 'Cause whether or not your artists or your clients wanna participate in AI, you're doing them a disservice if you're not actively learning about what's out there and what's available.

 so you know, hats off to you guys for already taking on that second one, but if you haven't already, I would just really get to know your catalog and understand what it is that you control.

[00:38:39] Dmitri: Anne?

[00:38:39] Anne: Bridget stole my answer. Um

[00:38:41] Dmitri: Great. That's the fastest speed round ever.

[00:38:43] Anne: Yeah. What I would just say is like, obviously exploring tools like ElevenLabs and Stability AI, they obviously have a lot of really amazing programs and opportunities to work, and the first part is just having a conversation around what those opportunities are.

[00:38:57] Dmitri: Great. We only have one minute, but I only have one more question. At the top, I shared what Goldman Sachs thinks new tech could add to music. After everything we've talked about, what makes you believe that this time, technology can grow the pie for artists and rights holders instead of shrinking it?

Someone wanna help round us out by going first? What do you think is gonna help grow it this time, technology this time? Oh, yeah. That was what the whole panel was, Dmitri. Weren't you listening?

[00:39:26] Ana: I can go if you want. I, I don't think there's been a time where technology has shrunk the pie. I think technology has always grown the pie, and I think you just see that with every technological shift.

The thing that comes with that, though, is that roles change. And I don't think roles necessarily... There might be some that- aren't as important anymore, but a lot of new roles are added. And you could- there's report after report that shows that, and that's what I think is going to happen here, is the pie is gonna get bigger because it's just what happens in these type of, like, cycles.

[00:39:57] Bridget: Uh, I think we're gonna hear a lot more from a lot more people that we may, may not have had access to hear what their self-expression meant because these tools will help democratize creativity, in a way where people will have access to things that otherwise would have been costly or too costly for them or too hard to learn.

 and then from a technology perspective, that also goes and applies to folks who maybe who have lost their voice, um, and need restorative benefit of voice restoration. So, I think there's just gonna be a lot more, creativity and self-expression that we're gonna see, and that's gonna open a lot of doors.

[00:40:27] Dmitri: Nice. Anne, take us home.

[00:40:29] Anne: Yeah, I think, like, that is like, it's less around, you know, when new things come, there's a whole slew of education that needs to happen around it, but I think the opportunity is, amazing. Like Bridget said, as those who maybe are limited in the skills or the access that they've had before, it opens up a whole other opportunity of creativity that's been limited before.

[00:40:52] Dmitri: Amazing. I came up with this topic for this panel before I knew any of you or even talked to anybody at your companies, and I said, "This is what I want. I want licensed AI companies. I wanna deal with name, image, and likeness." I got all of that, and I got these veterans of both the labels and music and agency world and tech.

[00:41:10] Bridget: And an all-woman panel.

[00:41:11] Dmitri: An all wo- well, almost. Oh, yeah. But yes, that was, that was lucky. Yeah, let's, let's hear it for all of you. Thank you so much, Ana, Bridgette, Anne.

[00:41:20] Bridget: Thank you,

[00:41:20] Dmitri: Shure. And do, if you haven't checked out the Stability AI, if y- after talking about having this whole intellectual conversation, go spin the wheel.

They set up a Music Tectonics specific, activation of how it works with stable audio. You press a button, it spins wheels just like at the carousel. It feels like we're at a fair, and you start playing music. I did it, and then I, there was a keyboard there. I don't play the keyboards. All of a sudden, I got this bass track, you know, of, of everything and started...

I was like, "Holy cow, I can do this." I, it ac- it sounded good, actually. So check that out.

Again, thank you so much, guys. This has been great. Appreciate it. Thank you.



Let us know what you think! Find us on LinkedIn, and Instagram, or connect with podcast host Dmitri Vietze on LinkedIn.


The Music Tectonics podcast goes beneath the surface of the music industry to explore how technology is changing the way business gets done. Weekly episodes include interviews with music tech movers & shakers, deep dives into seismic shifts, and more.

bottom of page