Making Sense of Music Streaming

The rise of streaming platforms like Spotify has reshaped music economics, revealing the industry’s longstanding flawed incentive structures that can make it difficult for artists to be paid for their work. These issues are only further exacerbated by the rise of AI, which raises questions around originality and copyright. In this ever-evolving industry, how can we better protect artists and ensure the longevity of the music we love?

To sort through the complications of the multiple models, I sat down with Mike Huppe, the President and CEO of SoundExchange.

Below is a lightly edited and abridged transcript of our discussion. You can listen to this and other episodes of Explain to Shane on AEI.org and subscribe via your preferred listening platform. If you enjoyed this episode, leave us a review, and tell your friends and colleagues to tune in.

Shane Tews: I have been so befuddled by how people in the music industry are paid for the work they do, especially artists. So we’re going to dive right in. I’m a big Spotify listener, there was a time in my life, I was a big Apple music listener, and there people who get their music from Pandora or YouTube or the radio or another medium. And depending on what we use, the artists all get paid differently. Who thinks that’s a good idea?

Mike Huppe: That is correct. The way to think about the music industry is like the game of baseball where you need to make a ball and a bat to play the game, but they’re completely separate products with different characteristics, different P&Ls, and different companies. The music industry is kind of the same: you have the song—so the notes, the lyrics, the tablature, the chord progressions—as one product and you take that song and you have somebody perform it and turn it into a sound recording, and that’s a separate product. For almost everything in the music industry, you need permission to use those two things, but you go to different people and pay different people to get them. That’s part of it the heart of what a lot of the complexity is.

For instance, on FM radio, when you hear a recording, the songwriter gets paid. The radio has to pay for that one part of the product, but they don’t have to pay for the sound recording. That’s a huge aberration around the world. In the United States, when the radio plays RESPECT by Aretha Franklin, FM radio only has to pay Otis Redding (who wrote the song) or his estate. They don’t have to pay Aretha Franklin.

Is this just in the United States?

It is specifically in the United States. I would argue it is one of the greatest inequities in the music industry. It costs American artists money, not only here, but also overseas. Since our radio doesn’t pay performing artists, many overseas countries use that as an excuse to not pay Americans. So for instance, if you’re sitting at a cafe in Paris, and you hear Bruce Springsteen come over, there’s a royalty that gets paid, but it never works its way to Bruce Springsteen.

This is something SoundExchange and the whole industry has been working on for a long time. We have legislation in Congress right now called the American Music Fairness Act, which is basically an attempt to level the playing field and get radio to pay for both parts of the music: pay the songwriter as they do, but also pay the performer because it’s the performer that brings the song to life.

Why is FM radio different from a legacy perspective?

Well, initially, there was a fight to get FM to pay. The radio industry in the US grew up a little earlier than the recording industry. Radio started to take off in the 1920s and 1930s, and if they wanted to perform music, it would have to be live because they didn’t really have records. So, they would bring in a band or tiny orchestra and they’d stand around a mic coming down from the ceiling and the radio company would pay them every week to come in and play. For example, Ford Company, as part of a radio program, would pay this one band $13,500 every week to come in and they would broadcast it out. Then all of a sudden, new technology brought along these little black discs (records) that you could buy for 75 cents. And radio said, wait a minute, I could buy that disc for 75 cents rather than paying a band to come in and play. Performing artists back then recorded albums with the expectation that radio would not be allowed to just take and broadcast that album.

In the 1930s, this issue was brought to the courts and there were cases that went both ways. But ultimately, the law shook out that there was no performance right. It became accepted that federal law does not have a performance rate for sound recordings for FM radio.

There are different policies, though. Take my example of RESPECT by Aretha Franklin. When you hear that on FM radio, a royalty goes to Otis Redding but not to Aretha Franklin. Now you start streaming the same song from your phone. Now it’s a digital transmission. All of a sudden, Aretha does get paid. That’s completely unfair. You have a $12 billion FM radio industry that is making its money off the music, and they’re not paying the recording artists that create their main input.

Let’s move to the next segment of challenges, which is when we introduce artificial intelligence into this. Now someone can use a song written by a specific songwriter and plug another artists’ voice into it. How do you protect artists when it comes to this and other AI challenges?

It is mind numbing how quickly artificial intelligence it has taken hold. It was only about 15 months ago that Chat GPT first entered the common parlance and now you can hardly have a conversation like this where AI is not a topic. I’ve never seen a topic take hold of public discussion as rapidly as this one.

To protect artists, you have to respect the creators that form the foundation of many of these large language models, respect what they contribute, and make sure that they participate properly. AI has really interesting, positive things that can do for the music industry, but at the end of the day, these models are trained on huge swaths of copyrighted content. They scrape all these sound recordings and photographs that are copyrighted and protected. We think it’s pretty clear that if you’re going to go and copy all of those for a commercial purpose, the way to protect creators is to respect them and bring them into the process.

I talk about the three C’s: consent, credit and compensation. It’s okay to use creative works and use other people’s songs or recordings or pictures or photos. But you need to get permission to do that and have them share in in the wealth that you’re going to create. It’s only fair that when AI companies build their learning models and their algorithms using copyrighted content, they allow creators to participate in some of that revenue at a fair point. All companies have a right to try to make a profit and earn money on their own, but you can’t leave the creators in the dust. If no Creator can count on being able to make a living off of their creations, we’re not going to have any more creations.

What frameworks exist to protect copyright when it comes to this? Is there a point where it’s considered unlawful? Or is it just unfortunate?

All of this is being fought in the courts right now and but it’s going to be several years until it’s all worked out. I have my own personal views, not only as a lawyer, but given what I do in the music industry. Is it lawful to scrape copyrighted content to create a commercial database that you then monetize? We would say no.

There are also issues about “right of publicity.” You’ve no doubt heard about some of the AI sound-alike fakes that happened—The Weeknd and Drake had a very famous one a few months back. That’s not just about copyright, but also about what I would call a right of publicity. It’s not just about singing, it’s about the persona that they create. If you’re creating a fake Drake track, you’re doing that to try to capitalize on the Drake brand and capitalize on Drake’s listeners that he has built up over years. There are actually two pieces of legislation in Congress right now to create a federal right that isn’t even copyright, but it has to do with protecting a performers’ voice or image or likeness.

There are other things that can be done to protect artists, though. I would love if AI companies would lean into the creative process. How about they recognize what artists and other creators bring to the table and work with them rather than against them? You see other proposals in Congress to help make sure we have responsible AI. So things like, do they need to put a marker on things that have been created by AI? Label that it’s an AI creation, as opposed to human creation? You could imagine a world where we have people who gravitate toward AI created work and we have other people who want human created things. Perhaps there should be requirements around saving all the data and metadata associated with a scrape, so that when we finalize how we can reward creators, there’s a method to do that. There are also recommendations in Congress about putting some requirements on AI companies where they have to self-report on privacy and some of the dangerous things that AI can do. That way, we could have a reporting structure where there is some regulation so that we have responsible actors in the AI companies.