Building AI for Music Without Breaking Trust

Where does AI stand in the music industry as we move through 2026? In this conversation, Dmitri talks with Mike Pelczynski, Chief Strategy and Impact Officer at Voice-Swap, about the evolution from fear to curiosity around AI in music, and why building ethical AI systems from the ground up creates long-term value. 

Mike shares about how his work architecting fan-powered royalties at SoundCloud connects to his current mission of building fair, creator-centric AI voice models. We explore VoiceSwap’s approach to licensing, attribution, and ongoing royalties for artists whose voices power AI models, and discuss what happens if the broader marketing doesn’t prioritize fairness. Mike also introduces the concept of voice models as a “source of truth” for identity and ownership in an AI-powered future 

The news

• the musical self-expression spectrum

• Spotify eyes AI ‘derivatives’ as new revenue stream for artists – says its tech to let fans make remixes and covers is ready - Music Business Worldwide

• The music industry’s boiling frog moment

• New York Senator Wants to Ban Ticket Scalping Forever

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] Dmitri: Mike Pelczynski is here. He's Chief Strategy and Impact Officer at Voice Swap, an AI voice company focused on licensed content based voice models and enterprise infrastructure for the music and media industries. Hey, Mike.

[00:00:14] Mike: Hello? Hello. Hey, Dimitri. Hey everybody.

[00:00:17] Dmitri: A little context here. Voice Swap works directly with artists, rights holders, labels and platforms to build voice models trained from scratch under clear contractual and governance frameworks with artists participating in ongoing royalties and downstream value creation.

Cool stuff. Formerly, Mike was with. Warner Music Group and SoundCloud, and he also advises several other music tech startups and is on the cusp of new thinking across the industry, which is why we have him here. Mike was a principal architect of fan powered royalties at SoundCloud, the first user-centric streaming payout model adopted by a major DSP and widely cited as a catalyst reshaping streaming economics.

We've got a podcast about that. That was super fun and amazing project. Welcome, Mike. It's great to have you here.

[00:01:02] Mike: Well, thank you. Thank you. Thank you for having me.

 since this is being recorded, I guess I have to swear less.

[00:01:07] Dmitri: No, it's fine. I don't, it's all good. So let's dive in.

I have a few questions for you. and uh, yeah, we'll just take it from there. Coming out of 2025, where are we with artificial intelligence in the music industry? Mike

[00:01:20] Mike: I'd say to overly generalize, we went from fear. to concern, and now I'd say it's at the stage of curiosity for music in the industry. the other industries outside of music, we, we work with, say the ad market and other adjacent markets, they've moved on past curiosity to kind of go into experimentation.

But, um, I'm pretty happy to see that most of music sectors may be generative music. Or the voice economy. it might, you know, most of it's happening in the background or in private. There's a lot of private models going around and experimentation, but it's still, it's functional maybe is another way of putting it.

 people are putting their toes in it and trying to actually learn how to swim. but yeah, there hasn't been a big plunge yet to be completely frank.

[00:02:09] Dmitri: Yeah, that makes sense. So your, your past work on user-centric payouts in streaming put forward this fair model of payouts for artists that's based on listening rather than on market share or backroom negotiations.

How does that thread continue into your work now? 'cause it's, I mean, you were at a stream service, now you're at an AI voice company. I'm curious, how does that all connect in your mind?

[00:02:31] Mike: the very upfront. Answer to that is it was a really difficult task to repair the past. and it was one of those efforts that we saw through data, we thought through, building the, the solution to streaming before we had the AI conundrum, as we were seeing the beginning of fraudulent streaming, as we were seeing this kind of convergence.

The signals of this convergence that we're now, we were trying to repair the past, evolve the economic model, and then some, which took a lot of different kinds of effort, right? It wasn't just building trust in what we were creating and putting into the market. It was also getting everyone and every stakeholder to, to admit there needs to be a change.

And then the most difficult part was getting them to adopt the change when they were used to. The way things are being done and the way money is being, the value of playlists and things of that nature. So this, on the other hand, was and is a huge opportunity at shaping this at the genesis of, let's call it the next format of music.

Some, some people call that, right? It's not necessarily another format. If this technology now creates multiple opportunities and challenges. some of them is new revenue. Some of them is new ip, especially in the voice section side of things. but yeah, I'd say it was really coming out of that experience in the trenches of streaming and seeing this new frontier and seeing the vision and the runway to create.

Something that unilaterally can benefit all if done. Right. So yeah, that was the initial appeal three-ish years ago to jump in and start building in this space.

[00:04:12] Dmitri: Yeah. Yeah. So on the stuff you were doing at SoundCloud with the user-centric payouts, it was sort of like, how do we fix some things, adjust to an equitable payout method

And what you're seeing now is more like, how do I be part of something from the start, from when it's first emerging in the market? Like how do we make this piece of it fair? Um, and both are creator centric too. I mean, that's the other piece of it that's interesting, right? Like SoundClouds.

Brand has been like the creator first streaming service anyway, like the artist, you know, it's oriented around the, the artist just uploading and sharing stuff, finding collaboration opportunities, that sort of thing. and the model that voice Swap has seems to be very creator centric as well, both in terms of creating tools for creators to start playing with AI voice, but also treating the, training data creators as creators.

And having attribution for them as well as, payout and, and royalties there.

[00:05:06] Mike: Absolutely. It's actually, we took a good amount of time in 2023 before launching publicly to answer that question of how would we monetize this? How do the royalties work that we not only are doing it right, but we're also monetizing something that you couldn't have monetized before.

So the fact that. People that are subscribing to our online service of, with our public models. So all those creators and artists, and they're all legitimate artists. there's no such thing as Jazz Singer, A and R&B, singer B or anything like that. It's legitimate artists. and some of them quite known that have their models public.

And people are using them throughout their workflow. So even if someone is using Liam Bailey's voice or one that I can't announce yet, but this year we're releasing, which I'm sure most people on this call will know. They're using them in their workflow and in in that process of just swapping a voice, trying out anything, experimenting, and it's not even gonna end in a commercial release.

It's just going through the workflow to pitch even to those artists if they want to feature on it, they're all making money during that process. So even. even through a demo of just using and, and trying things out, artists are generating passive income through the technology, which was, that was the initial impetus to get people to adopt this because they, the initial class of artists that we worked with, they not only trusted us to try to do this in the right way, as other companies were blitzscaling and using, you know, questionable.

Efforts and models. We did it in a way to say, let's do this together. Let's grow, let's get you money. And you could see how this is growing as we build out all the other avenues and potential for their voice models that they own, that they take ownership of and have control with us. So it was the bedrock of doing this, right?

That now those artists, they can go to any other platform, they can go to a You know what, some would consider a competitor of ours in the voice space in music, but they can through API. Use our, their model with us, with other platforms, other companies, other enterprises. So really the way we, we envisioned it and have deployed it is.

It's not just sitting with us, it's us partnering with a voice, talent, maybe a voiceover actor, a musician, artist, singer, producer, and then having them have the ability to go out into any ecosystem and to leverage that, what we call source of truth, which is their voice model. but yeah, the revenue, the royalties, that was the bedrock of where we started.

Everything to, from that point is where we started to build everything else. and quite frankly, experiment in some respects.

[00:07:47] Dmitri: And it is in, differentiation from a lot of the generative AI models that are out there who have come out saying. Licensing training data is, is an expense that would kill this and have even had investors come out and say, it doesn't make sense to, to license or deal with attribution or anything like that, because that holds the technology back.

That holds the shifts, the evolution in how people are. Creating content back, et cetera. what's the fair path forward? In a way, I feel like voice swap, what you're saying is voice swap is a model for a fair path forward, certainly in the, in the music industry. and so I'm curious if we could talk broadly about that and then what happens if it doesn't play out that way.

[00:08:31] Mike: So broadly and and part of my dual role at the company my, in my impact role, I'm heavily involved in legislative bodies policy shaping, um, namely in the uk. Uh. With the, the Creative Rights and AI coalition out there in the UK and working with Parliament and with Wipo, and the reason I'm mentioning these institutions and bodies is going back to your question, we decided as a founding team.

We're gonna go this strategic long road to do this in a responsible way because a year, two years, three years from that decision, we knew that value would come back to us because we wanted to take these artists, these this voice talent, to work it in music, and then give them the ability to work outside of music.

And in order to do that. Our base model had to be proprietary. The training that had to be clear lines of who owned what. and in that instance, that opened our business and our company and our partners, the artists, to be able to work with the advertising markets and other markets. So we also have text to speech capabilities and things of that nature.

So. that really opened up the avenue to do bigger business with trust. And the trust is baked into the way everything is built and our agreements with artists. So, a long-winded answer to say there is huge value in this space to do it ethically, but you have to have the commitment from the team and everyone involved and investors to really dig your heels into that and stick that road to the point that, you know, we've been around for.

Just shy of three years and the benefit of going that route, we just started seeing that come back to us in the last, I'd say, six to nine months, where you know, people going to other companies and saying, you know, word of mouth or through other means that they came to us from other potential companies saying, you know, the way you're doing this or the way you're going about this, makes us much more comfortable to do business and work with your talent.

Yeah.

[00:10:36] Dmitri: Yeah, that makes sense. What, I guess the, the second half of the question is, what happens if, the fairness approach, doesn't play out in the larger tech and society and engagement in the market, what happens then?

[00:10:51] Mike: It's, everyone's always at the mercy of that, right? Even if. There's these large generative platforms and they strike deals with larger rights holders or you know, big independents and they set a presence that is in a gray area.

Right? that's something that definitely who has control over that. those involve the decision makers, especially with larger bodies that have the potential to, you know, help inform those decisions. But yeah, no one's bulletproof because it could easily be. We survived the Blitzscaling era where there were other platforms, blitzscaling and having crazy voices on there that some were like, you know, hate speech inducing and crazy, just crazy.

I'm not gonna say it 'cause we're recording this, but, uh, we survived that era. And to your point right now is it's just getting to the wire, it's getting down to the line of, you know, now people monetizing this or figuring how to utilize this technology Even if it goes in that route where say the ethical or responsible approach is skirted by other companies, there's still the concept of ownership.

And that's the thing that we're really sticking hard to, which is everyone on this call, you know, and however long in the future, and I really do believe this, all of us are gonna have a voice model of our own, the way we store pictures and paperwork and things on Dropbox. Like it's something that inherently we're going to have to have that definitive source of truth of.

Who we are and what we sound like, however far in the future that is. So there is still a need for ownership. There is still a need for utility as this technology entices people to use it. But the part of it is like, going back to that frame of curiosity. Someone could have curiosity to try the technology, but then they're, you know, as they take two steps forward, they'll take three back because they don't feel like they have control over how it's being used or where it lives.

[00:12:47] Dmitri: Mm-hmm.

[00:12:47] Mike: Or who's the custodian of, you know, their intellectual property or, or even just their identity. So that really, I think, is this anchor that no matter where those. Areas go when it comes to, you know, ethical, non-ethical developments that is still an anchor that everyone can agree to that still find value in it.

Because like, I'll, I'll give an example. There's a major, label Country star and they were really interested in. Getting into this technology, but they needed to figure out the initial utility. Their team was like, we already know how we would use this. We need a text to speech model private so that their team can do the hundreds of radio spots that that artist, that talent has to do.

So that was that one form of utility that they knew willing to pay for. Invest in the time to do it. And then once they had it, you know, they did two birds with one stone. They have this functional thing that solves a problem. They have this tool for other things. But at the same time, they also created a source of truth for this, you know, well-known artist that their representatives in legal teams, if they find something nefarious on another platform, on a YouTube, TikTok, whatever it may be.

 we also operate as the rights management. Agency on their behalf too, helping in terms of enforcing that, that, right. So it, it goes back to that anchor without going too much detail, that will always fortify and push back against the potential irresponsibility of other companies, is my belief. But you never know.

Is my asterisk.

[00:14:21] Dmitri: So, so in a way you're saying like this is a differentiator for you. If, if the market goes a different direction, then it's just people can choose to work with companies that have put their flag in the, in the sand around fairness and Attribution and

[00:14:37] Mike: mm-hmm.

[00:14:37] Dmitri: Payouts and all that kind of stuff.

Cool. Um, Mike, I've got one last question for you, just expanding out, because I know you're a big picture thinker based on our past conversations and, I I don't wanna focus entirely on AI or voice necessarily. So outside of that space, which we've got got a chance to talk about, what other trends or innovations are you keeping an eye on in 2026?

[00:14:57] Mike: I'm a bit obsessed with admin tools. I have been over the last couple of years.

[00:15:02] Dmitri: Card a

admin tools.

[00:15:04] Mike: Yeah. It's so not cool, but it's so goddamn cool. Um, because, you know, most of, I'm sure most of people on this call have seen it or have heard stories from colleagues or former colleagues where it's, you know, artists being dropped from labels, people finding teams, you know, the usual story of.

Someone finds success and then they just have to assemble a team around them, and then they start to learn how to operate as a team. So I just, I've been really interested in watching what happens in the space, you know, with these companies like Mogul rollout, you know, all these kind of companies that are trying to solve this, this problem that.

Yeah, if there's these teams of artists and there are people around them, if they can organize themselves a little bit better, you know, honestly budget themselves a little bit better, you know, we'd have, they could push back. I mean, it's, it's an interesting business, I think, potential business, but I don't feel like there's enough chatter around it.

But yeah. That's the very left field one to go.

[00:16:02] Dmitri: Yeah. No, we've kind of seen that kind of percolate. I know water and music talked about the backend is sexy again or so, something like that when they did a

[00:16:11] Mike: Oh

yeah.

[00:16:11] Dmitri: A survey a little bit. Maybe it was last year. I can't remember. But, but the whole idea that like, there's all these problems that still need to be solved and it might not be sexy on the front end, but what's happening on the backend really changes like financially has a substantive impact on the industry, on individual companies, on artists, on, partners and, and things like that.

So totally get that. Awesome. Mike, thanks for letting me put you in the hot seat on this and having that conversation.

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.