Skip to main content

SC-070 · Expert

The Neuroscience of Music: Exploring AI, Culture, and the Future of Sound

Guest: Tom Fritz, Research group leader at the Max Planck Institute for Cognitive and Brain Sciences

Summary

Tom Fritz leads a research group at the Max Planck Institute for Cognitive and Brain Sciences that studies how music can aid rehabilitation and fitness. His answer to the AI question is that the body processes an AI track and a human track of equal quality the same way. What separates them is what the listener knows about the artist and how much agency the listener has.

His evidence comes from fMRI work on how liking or disliking music changes processing, and from field research with the Mafa in northern Cameroon. The Mafa recognized happiness, sadness and fear in Western music they had never heard, which points to universals that may still work for stroke patients. He also found that physical effort and musical expression amplify each other, which led his lab to a technology that lets people play music through the effort of a workout.

Fritz splits the value of music roughly 50-50 between physiology and cultural context. He explains why AI music can feel soulless: genuine emotion drives motor control in singing and playing, and listeners read it as an honest signal. He predicts live music will never die out, and that AI will pressure the industry to make licensing easier.

As of the episode's release on 27 August 2024.

Key takeaways

  1. 01On a purely physiological level, listeners who know nothing about the artist will respond to comparable AI-generated and human-made music the same way.
  2. 02Knowing an artist's story, even small details of their life, changes how the brain amplifies or filters the music, which makes storytelling a lasting advantage.
  3. 03Agency drives attachment: DJing, playing an instrument, a detailed AI prompt, or even stepping in time with a beat while running all deepen identification with music.
  4. 04Fritz explains the soulless complaint through honest signalling, since genuine emotion controls voice and muscles in ways that are hard to fake in a studio.
  5. 05He suggests part of the demand for AI music comes from licensing so hard to arrange that even willing buyers cannot get a track.

Guest

Questions this episode answers

Does AI-generated music affect the brain differently than human-made music?

Fritz says that on a purely physiological level, a listener who knows nothing about who or what made a track responds to comparable AI-generated and human-made music in essentially the same way. What changes the response is not the source of the sound itself but what the listener knows or believes about it.

What did research with the Mafa people in Cameroon reveal about music and emotion?

Fritz's field tests with the Mafa in northern Cameroon found that people there could recognize happiness, sadness and fear in Western music they had never been exposed to before. That result points to some emotional responses to music being universal rather than something a listener has to be culturally taught.

Why does having agency over music change how much people connect with it?

Fritz says agency, whether from being a DJ, playing an instrument, writing a detailed AI prompt, or simply moving in time with a beat, deepens a listener's identification with a piece of music. The more control or active participation someone has, the stronger their emotional connection tends to become.

Why do some listeners say AI-generated music feels soulless?

Fritz explains it through honest signaling: genuine emotion drives fine motor control in a human singer's voice or a musician's playing in ways that are difficult to fake, so listeners pick up on that authenticity. He says he understands the soulless complaint even while thinking the underlying physiological response can be similar either way.

Why are people turning to AI-generated music instead of licensing real songs?

Fritz suggests part of the demand for AI-generated music comes from how difficult and slow it still is to license an existing track for a specific use, even when someone is willing to pay for it. Generating a new track with AI sidesteps that entire licensing process.

I have a little prediction that no matter how much AI we have, the act of making live music will probably never die out exactly for this reason, because there's this human and you can relate to them.
Tom Fritz

Episode notes

Music and neuroscience converge in this episode with Tom Fritz from the Max Planck Institute, where he explores how AI and cultural factors shape our perception of sound. Drawing from his innovative research, Tom highlights the transformative role of music in health, fitness, and rehabilitation, emphasizing its universal impact across cultures and its potential to improve lives through science and technology.

Highlights:

  • Exploring the link between music and health.
  • Understanding the effect of AI on music perception.
  • Cross-cultural insights into the emotional power of music.
  • Utilizing music for rehabilitation and wellness.

Topics

Transcript

Transcribed automatically. Names and terms may be misspelled.

Read the full transcript

So I have a little prediction, you know, no matter how much AI we have, the act of making live music will probably never die out.

Exactly for this reason, because there's this human and you can relate to them. This will play a huge role also differentiating human-made music from machine-made music. But the more AI-generated music we'll have, probably the more important this will become in future and the more of an extra value this will be, you know, to know, this is actually, this is real. I meet young producers all the time that are scared about AI.

What's going to happen with their career in the future. And in reality, I don't really know, but who better to talk with than Tom Fritz, a researcher in neurology that understands the connection between your brain and music. And we talk about what does it mean to listen to music, what happens and what doesn't happen. Hey guys, and welcome back to the Sound Connections podcast. Today I'm speaking with Tom Fritz. Tom, you me met over a phone call a few weeks ago and we talked about what you work with, which is really, really interesting.

And we also went into AI. But you sort of work in the intersection between music and health. And I would love to have this episode to understand sort of your findings and what you work with. And then also pivot that to a part of the conversation we had last week, what does this mean for AI music? But let's start about understanding who you are and what you do. Yeah. Hey, hello everyone. So I'm Tongtritz.

I have this little research group at the Max Planck Institute for Cognitive and Brain Sciences. It's called Music Evoked Brain Lasticity. And what we try to do is we try to optimize the use of music in rehabilitation, but more recently also in fitness. And you told me that you sort of

Haven't really come from music per se and got that interest at one point. Can you walk me through the journey of you being where you are now? Well, it's more the other way around that I hadn't come from sports. So now I'm invited to give these lectures at these sports conferences, this neuroscience of sports, for example, conference. And I really got interested

In this whole sports domain through the music. So I had been doing music ethnological field research. Yes, that's true. Where I had been investigating basically how people reach musical euphoria very effectively through a combination of physical exertion and quite minimalistic music making.

Okay. Interesting. What does that mean? So, when I was doing my, my PhD thesis, I was working with these super high tech machine. You know, with, with this, fMRI, these machines, cost millions of euros. Yeah. So,

We have several three Tesla machines and seven Tesla machine and you usually say each Tesla costs one million euros. So I was working with this super high tech and I was looking at the auditory pathway and how your attitude towards the music. So if you like the music or if you dislike the music, what effect that has.

Or the actual physiological processes of processing the music. That was very interesting because it has enormous effects and is fascinating. But I was asking myself, so I'm looking at this processing of this music, but what I was doing is I was...

Presenting music and I was presenting manipulated versions of this music that were manipulated in their acoustic parameters such that it would sound more unpleasant. Yeah. So you'd hear three versions of the same music and three different pitches that were selected to be, you know, a really nasty combination so that every sound that you hear is a, you know,

Is a terrible experience in itself. And we were presenting music backwards and we were doing a combination of both. So that was quite effective at, you know, investigating these networks. But I was asking myself, so how can it be then that, you know, I love to listen to somewhat something and my neighbor hates to listen to exactly the same thing, even if they had grown up in the same village?

So what's going on there? And I was applying at that time for some funding for a music, ethnological field research project and I got it. And so I went off to visit the MAFA in Northern Cameroon. And with them, I was doing a battery of music experiments, you know, to learn more about, you know, what's

What has to be culturally learned when we listen to music and what is cross-culturally comparable in the psychology of how we perceive music. But then, in the first weeks when I was there and nobody wanted to participate in my experiments because they first wanted to get to know me, if I'm maybe a dangerous guy, yeah. I mean, in the end I can say, you know, happy end, so they concluded I'm a harmless nerd.

And so everyone was totally unconcerned in participating. But that was after I was participating in their musical rituals. So that was part of the reason I was there is I was interested in, what's actually the common denominator of what they think is music and what I would think is music. And that was very, very interesting and enriching. And that also led me

My insights that led me to Jimin, to how physical exertion or even mental exertion would combine with music making. Well, there's so much to dissect here. So let's try to take it part by part. So the first part of your studies you did, what were the findings and conclusions that you would like to share that maybe could inform the music industry of some really interesting stuff?

Now you have to specify which research exactly, because there's a whole range of stuff I had been doing before the MAFA study already. But of course, the MAFA and also these were quite interesting findings that also put my little lab on the map in a way is that where we found that there are actually certain, you know,

There seem to be some universals of music perception that are also quite interesting in this rehabilitation domain because that also means that, you know, when you have a stroke, for example, it usually affects mostly the neurology that has been formed through

Through cultural experience, know, so, you know, all these synapses get lost and et cetera. Yeah. And so if you have something in music where you know, there are certain mechanisms that, you know, that, that should be universally perceived, there's a pretty good chance that will also still work with your patients. So what are, what are those things that you

Found to be universally true in music.

Well, that's a bit of a longer story. It also reaches into, what's so different about their music and our music? And that's quite fascinating, I tell you, because it seems that some of the things that are universally perceived are not necessarily actually universally available in music. So...

Music may not require these universal processes. For example, the music of the mafa, it does not require emotional expression like happiness, sadness, scariness, know, although

In my tests I see that the people there can actually recognize all of this from music that they have never before heard in their life. Although, I mean, you know, it's science, so you know, the truth is always a spectrum. And there are people who are very good at recognizing this and there are people who are not so good at recognizing this. Yeah. But overall you can say that, well, people are actually, you know, they would

Listen to a certain musical piece and then they would be successfully matching this to a facial expression. So that's quite amazing actually. But then you know the Mafa music it doesn't even require these emotional expressions because in their music these culturally learned aspects they are very strong you know and if you and you don't need

These emotional expressions, obviously, to get very emotional. So that was also interesting because, you know, they listen to some of these recordings of their music and they begin to cry because they are so emotionally touched and they find it so beautiful. But it's also, I mean, to a Westerner, that's, you know, not necessarily the most beautiful sounds, but if you have very strong

Experiences that you associate with the music. That's what can happen. And that's actually also where those other findings also relate to that I have made that led me to these, you know, to the Jimin invention is that especially if it's very strong physical experiences that

That you relate the music with. And very, very interesting things happened. We found that there's a... So in all my experiments since then, we found that there's a positive interaction effect, that's what we call it, between the experience of effort and being musically expressive.

What does that mean? That means that these two processes, they benefit each other in our perception. So if you are musically expressive, but you feel that you are doing this with a lot of effort, you tend to interpret this as a, you know,

As Leidenschaft, is what we say in Germany. So as something positively emotional. The reason for this is that this physical arousal that you have, for example, when you're trying to do a hard thing, like working out on fitness equipment.

That's exactly what we do, for example, with this Jimin technology that we're bringing out to the market. So basically what you're saying is that the surrounding circumstances of what you do and where you are when you listen to music could positively affect your experience with the music. Yes, that's always the case. But what we observe is that when it

Comes to changing all kinds of parameters, cognitive parameters, and also physiological parameters, you want to optimize these environmental parameters because that happens mostly in a sweet spot.

I want to understand a bit more because sort of where with the conversation is heading is also understanding the impact of AI music, AI genetic music, how that's going to affect listeners, what sort of things we need to know about, let's say the universal truths of music and how you perceive it. So let me just ask a question. So what are the things that the music industry is very concerned about right now is how big a part AI genetic music will have in the

Let's say the music streaming environment, like there's a lot of passive listening and playlists. How much market share will AI take? But before we can sort of really discuss that in a qualified manner, we need to know what is it that makes people enjoy listening to music? So in the case of, let's say we have a very well-generated track from AI that maybe from A to C is a good, calm singer songwriter track. That's put side by side for human-made, traditionally made music.

Are there any things that you comment on those placements together? Would there be any negative or positive associations with music? Would people know? I just need to understand that.

Yeah, I mean, what we know from the experiments with the MAFA for example, is that, you do have a huge influence of unculturation, know, so cultural experience. And, and of course this will also play a huge role. Like, you know, it plays a role that people know.

Something about the artists they listen to. On the other hand, there's a number of mechanisms how music works just on the physiological level really, you know? And of course, the effects of music are interplay between both. But I tell you, you know, so that's what usually people perceive as immersion.

But from my, so one of the realizations I had from my also music, ethnological experiences is that probably the most,

The most cross-culturally valid universal of music is making music together. So that's where being yourself in the act of music making, because people will immediately build up this, you know, the context of

Or what the music actually means. It seems that we're doing this, we call it this semantic, the content of the music, it's loaded with this meaning very, very fast. We're in a situation, and you take two people,

People from two different cultures and they're playing two instruments from totally different backgrounds and they will still be able to have this musical, you know, dialogue. And although it may not be that you understand all the words, it is that you, you know, you create these situations together and these situations and the atmosphere in these situations, they will have, they will also give meaning

To what you do together. Okay, that's very interesting. So I hear three things right now, and this is just my unindicated dissection of what you said. There's a physiological effect of music, there's a cultural understanding that gives value to music, and then there's interplay between creating music. Correct. Let me sort of ask question by question because that's gonna help me. So the physiological role of music, if you see it isolated from...

Culture and the communal aspect is basically that there's audio waves that has harmonics that has all these kinds of things that do something to the human brain. Let's say in a hypothetical scenario where - Maybe not only the brain, but through the whole nervous system, yeah. -hmm. In a context where an AI-generated track would be comparable to human-generated tracking quality, vibe, style.

All these kind of parameters only from a physiological standpoint. Do you believe there would be a difference between what was human created and AI created? No.

So simple as that. I think on that level...

You know, it's the same things, it's the same procedure that's happening. That's really interesting. Could you dissect for me sort of a few of the things that the body and the mind reacts to when comes to music, like such as harmonics, tempo, beat, just so you understand sort of what is it that we receive that we can recognize and perceive as value?

I mean, basically the whole acoustic structure of the music is, you know, it doesn't matter from where it comes. Yeah. It's just, that's what's being processed and it's going to be processed in a similar way if it's as one source or another. I mean, there's a, I mean, it is more.

Interesting than that is that this information, depending if on your attitude towards it, it changes a lot and how it's going to be processed. So for example, in my experiments, in these physiological experiments, I saw that there are actual physiological mechanisms

In place to either amplify how this arrives in our auditory cortex or how to filter it out. So, you know, the knowledge about the music will play a huge role, but you know, the, like the baseline signal is just the baseline signal. That's interesting. So, so in a scenario, for example, where let's say

I would be traveling on the train, I'd put on a playlist on Spotify that I don't really have a relationship to. There might be songs I've heard before, but most of them are songs I haven't heard before. And I'm not looking at my phone. I don't know what Asus is playing or anything. I'm just listening to music. And I have these tracks come by. Basically what I hear you saying is we wouldn't necessarily, first of all, recognize this AI-generated music and we wouldn't devalue it or give it more value. It'd be of...

Neutral value compared to the quality of the context of which it operates in. I mean, if you know something about the artists that you're listening to, that can play a very important role, but another role can be if you are somehow yourself involved in manipulating the music yourself.

Because then the agency, it's with you. But if none of those two elements are present, then we can almost conclude that these pieces of musical works, given they have the same quality in whatever that means, would be of equal value to listener. Yeah. If you know nothing about the artists that you are listening to and you compare this...

Without you know nothing about the AI-generated music you're listening to. That will be the same effect. I think this is really, really interesting because one of the working theories that I sort of work towards and a lot of my colleagues work towards is we believe that the interaction between fans, communities and artists is sort of the future of how to keep the music industry healthy.

Because that bond, that cultural context is really what gives value to the transactions between these and how people use music and purchase music. It's already happening. You're It's happened before and well, fandom has always been present. But I think now what I'm trying to say, which is also what I've strengthened my belief in since you said this, is in order for us to really navigate through the immense amount of music,

That's gonna be generated by AI is really doubling down on the human connection, the cultural context, the storytelling. So break that down for me. Why and how does it matter with understanding what you're listening to? Knowing the artists or having a fan artist relationship, how does that affect us as human beings when we listen to music?

Yeah, I think the key is your identification with the music. So all these psychological effects that we investigate when we're using musical stimuli, I think the main key to all these psychological effects

Is something that, you know, is something we call self-efficacy. And this can happen at many levels. This identifying with an artist, for example, can do a lot with this, know, identifying with the musical signal, you know.

And this can be knowledge about the artists. It could be even information not related to the music at all. know, like this, you know, you know, the face she's in, you know, I'm, you know, that, you know, when, when she recorded this music, that's, you know, I can, I can relate to that. You know, like I had this, I know how she felt then. Yeah. And this, this does a lot. Yeah. But of course that's not.

Not all. mean in the end you know these this is just basic acoustics and a lot of it is still you know I mean you turn something faster you turn something slower you change something in the timbre you change something in the sequence of you know of harmonicity you know all this will

Will have some basic psychological effects on us. If you really want to focus on the music and you want to zoom onto the music, this will possibly require additional information. You're right, like this connection.

Between the artists and the listeners is going to be enormously relevant. And the most, so I have a little prediction that no matter how much AI we have, the act of making live music will probably never die out exactly for this reason, because there's this human and you can relate to them. But you know,

But of course it can happen at more abstract levels that people know actually through other sources and technologies how...

I mean, something about the intentions or, you know, or the life of the artists. And I think this will play a huge role also differentiating human-made music from machine-made music. Yeah. It also speaks to the point with a lot of, let's say independent artists that maybe are into the belief that as long as they make great music,

You know, I'll be successful. But Albert, now you're competing with like generators that can most likely do as good, if not better than you, if you're five years into your career, you're kind of independent artist. So the storytelling you bring, the community that you bring, the narrative that you have surrounded yourself with is probably going to be the differentiating thing in a career that allows you to at least have a

Chance of success is gonna be my assumption. Yeah, would, I think you're right. I think that this, mean,

I mean, music still speaks for itself and I do think that, you know, sometimes there's a certain magic to music and you hit a certain point where something, you know, resonates with people. But I think that, yeah, if you...

You know, like these elements, these additional elements, can help, they can help the listeners to, to focus in on the music. Yeah. I think that would probably in generally be a good, yeah, good suggestion to the artists, know, to, you know, how they deal with their communities. And I mean,

In the end, see a lot of that happening all the time is that, you know, it may not be enough to just make nice recordings and put them out there, but that this, you know, playing live, et cetera, it's also important in this storytelling procedure, you know, that this is actually, you know,

This is real. Is there a way to sort of, this is a difficult question that it might be weird, but is there a way to quantify the importance between the physiological response to music and the cultural understanding? Let's say that, again, this is my own math, try to meet to understand. So the physiological being pleased by the music you hear without context is like one, but then I...

Habit culture understanding is three. Like is there a way to have a correlation between the importance? I would say it's 50 -50. I would say it's 50 -50. Well, you know, 50 -50. So that's quite a lot really. It's not only the sound at all. Yeah. And, you know, and sometimes, you know, I mean, if you look at

these experiences that the Mafia have, know, mean, their music is not dying out for them, you know, although, you know, they may be exposed to other types of music here. Even the cool folks, you know, would actually travel to other villages. You know, they still do their own musical rituals and it's because, but then also, you know, this

This cultural information, it depends on what it is, that's of course what's making the difference then. Yeah. Okay. Well, that's somewhat encouraging, but also somewhat scary that, what I had to fill up naively, the story and the community aspect and cultural aspect is so much more important. So as long as you nail that, you're safe. But what you're saying is

Obviously it needs to be a really good combination of high quality music that people fully enjoy and then also having a story that people can attach themselves to. Yeah, I don't even know if it needs to be high quality music. mean, it depends where you see the quality. In the end, people may actually see that as, this is actually handmade. All of sudden as a certain quality.

into it, you know, and think, this is real. Yeah. And then, you know, quality is of course that's also quality. And, you know, maybe that's, mean, yeah, it has to be a fit with what people like to listen to. Yeah. It's really interesting because when I study music and I have a master's music production, but when I started my bachelor's, I was going into a class with very established people with the sort of the...

the concerto I went to, you know me, had a career before you get in. I was lucky I was young and I just wondered a lot. So I think I got in that way. But we were sitting in a class and we listening to our favorite music. And I grew up in a very religious household and I wasn't exposed to much music, which, you know, my conclusion is I wasn't exposed to a lot of high quality music. You know, I didn't, I didn't really listen to Beatles before I was 20, you know. So, so I was sort of very caught off from, from let's call it the taste makers. So

They were playing this very interesting music I've never heard before that pretty much everyone else has heard of. And I was playing, okay, this is my favorite music. And it was like some YouTubers that I followed that did like a home recording kind of thing. And for me, that was the best music in the world. And then I really compared to what they're doing. And I felt a bit embarrassed, like, okay, I can clearly hear that this is lower quality when it comes to the handcraft, the mixing, everything. And I hadn't really thought about that before. And I've been thinking about why is that, that I gave that so much value? But what you're talking about is

It doesn't need to be high quality. It doesn't need to compare to the craftsmanship of others. But if you have a story, which I've been following these YouTubers for years, that was my peak of this is quality music. And I think that's interesting. Yeah, yeah. And this type of storytelling aspect to it may become even more important in the future, you know, to...

follow up on what are people actually trying to achieve and identifying with these goals. Yeah. That's incredibly inspiring. And I think if you want to look at from an optimistic standpoint, when now an AI is probably going to disrupt a lot is there's some beauty in that we are forced to going back to creating these connections in these stores. You know, I, in my career, I work a lot with independent assets to upcoming assets that sort of come to midsize. And I find it very few and far between.

people that really have like a compelling, compelling story other than I'm emotional and I like to make music. Yeah. But you know, in the end, I don't think it always needs a very complex or anything story. It's just, you know, it can be something very simple. You know, it can be, this girl's singing for her grandpa and that could be enough, you know. But again, in my opinion, every single

powerful story as simple as it is, just needs to be clear. Like you, you as a storyteller need, you need to, in my opinion, need to be mysterious. All mysterious. All mysterious. you know, there's probably... there's in that as well. Yeah, exactly. know, and it's sometimes very good if it's not so clear. So, you know, I think it's the spectrum. And at the end, the art is, you know, arriving at the right spot.

in the spectra area and then finding this sweet spot where different dimensions, where you are in these different parameters, of the music, you arrive to be. And then there's this, sometimes this perceived magic happening where you don't know what the hell is it that kicks this song off and the world wants to listen to it. Yeah.

It's fascinating, but yeah, I don't think this will stop only because of AI. No, Okay. That's really interesting. Hugh, you have one point, which is sort of, I can't remember what words you used, but sort of the interplay with music being, being a part of the creation manipulation, which also speaks to some of the things that I've had a lot of discussions with the founders about. Like, okay, people can now generate music.

but is that going to provide them the level of satisfaction that are going to really, really attach them to the music or should there probably be a more, let's call it mastery experience that you have been a part of creating it, that you've had some more say than I press the button? Can you comment about that? How can people get that interplay satisfaction and connection? what's the limit? How much do you need to do in order to get that emotional connection?

So here, I think the key word is agency. And agency with music can happen on many levels. So if you're putting on the song as the DJ, you have the agency of, know, now that's the music we listen to and we try to party with. And if you switch the radio station, that's...

You've got the agency to change something. If you play a musical instrument, you have a lot of agency, you know, in a way, but it depends if you're improvising freely or if you're, you know, you're changing to the next song and you're playing this one. Well, then how you play it, you know, I mean, you have a lot of parameters you can vary, but, you know, I think...

This aspect of agency is enormously important. And if you're just listening to music passively on the radio desk, also already, you know, this agency, either you, it's triggered or it's not triggered. And for example, if you have previous experiences with music and you know, you know, that was when we were having this car trip.

And we were going on holiday and we were listening to this all the time. It also provides you with some type of agency. Sometimes it's even enough to have an illusion of agency. So for example, when you're running to music and you're synchronizing with the beat of the music, I think that's exactly what's happening is you have this illusion of an increased agency with the music.

that you know usually when you close your ears and you walk around you through bone conduction you hear

I mean, depending on how fast you walk, of course. But when you are synchronizing with the music, when you're running, instead of bop you hear...

and you have this you know what I like to call this illusion of a musical superpower you step down with your foot

Be instead of just duck, so, but it's, it's still this key agency. even if it's an illusion of agency, that can still be key to all this positive physiological and cognitive effects. Interesting. Okay. Let me go, let me go hyper -specific. So.

Do you believe that people will get a, generally speaking, a big amount of agency from, for example, generating music through platforms such as Suno and Udio?

Depends a little bit, yeah? I mean, if you feel that what you're imagining is being reflected in what you're getting, you know, that's, mean, that will probably then also, you know, if you, if something say, provide me with something happy, you know, of course that's not specific enough for you to feel that, yeah, that was exactly what I thought about. But if you say, hey,

You know, I would like to have something that sounds like happy birthday and I'm talking with Jakob today. It should include his name and maybe the name of, you know, his sister's brother, blah, blah. And, making the song about us going this afternoon to the park to celebrate his birthday and, you know, we'll pro

Vitamin with a bath of champagne and that's what the song should be about and it should sound a little bit like, you know, this and that song. And you know, and you then have an outcome, you probably feel, yeah, you know, I actually was involved in designing this, you know, and I can hear that. And it's reflecting all this information that I want in there and I thought of that, you know, and now it's there.

Then identification will be a million times higher, of course. I had a really interesting feeling like last week. So the context is I've been a music producer and a songwriter. So when I've been traveling, let's say around Europe and writing songs, sometimes you write a song that really resonates with you. Like I've written songs where the lyrics was really strong and the production was really strong. My vocal performance on that demo was really strong.

And I could get so excited and listen to it over and over again, 20, 30 times and still have this excitement. It might not be the best song in the world, but I would be so excited. And then I sort of went into exploring AI-generated music. And over the summer, I've been sort of building my new house and being taken care of the kids. In the evenings when I by myself was like, okay, I want to write some lyrics. So I wrote a lyrics, song, something that meant for me. And I started generating music. And the last four months, I would say, I've generated around 5,000 songs with AI.

And over the summer I made 12 songs, well 12 lyrics I made, you know, hundreds of versions and then I chose one and have it on playlist. And I got the same feeling. You know, I could listen to that song over and over again, like I wrote these lyrics or, but everything else was AI generated. Even the, actually the lyrics was structured using AI, but like was my little piece of that. And it gave me the exact same emotion. And I cannot differentiate it from when I was in the room, involved in every single step of process to me just.

Having a small piece of it, it's the same emotion. And for me, that's really interesting to think about because when I also hear the songs now, I don't feel ashamed at all. It's just, I'm as proud. I generated these songs with my intention. This was the Bible I was going for. There was a reference. I generated 150 songs to get this one song. I do feel like I put effort into it. Yeah, yeah. And that's interesting because

Basically, if I'm trying to combine my own thoughts on this, what you're saying is the physiological aspect of listening to music shouldn't necessarily differentiate whether or not this is human-created AI generated, given that they're comparable in quality and other metrics. And to my own experience, generating music does not necessarily bring less of the satisfaction from a creator's side. And then I was reading this comment on Facebook in a music industry group yesterday that I've read so many times. It's like people saying, yeah,

Shouldn't be scared of AI music because it's so less. And that's what people say all the time, it's so less. What does that actually mean? But probably what they're speaking about, which is my assumption, is that they're talking about they don't have a cultural relationship to that music. I understand this point because...

Because I think what music is a lot about also in the beginning of the creation of music as a cultural tool for mankind, I think it is in its roots.

A technology or a way to, you know, that relates to honesty of signaling. So, you know, if you have

If you have, and I can even tell you why this is happening on a physiological level. So if we are talking with each other and I'm smiling, you know, it's quite interesting because you cannot, you know, you cannot fake a smile. You know, there are muscles involved, the outer lower eye ring muscles that make it a real smile. And I can pull on it how I want.

And this is, if this is not going on, it's because this can only happen through so-called emotional motor control. Now the same thing is happening if we are talking and if we listen to each other's voices. So the prosody of it. Yeah. It's something if you, if we are in a speaker studio,

And we're pretending to talk to each other, that's a million times more difficult. Why is that? So yeah, maybe if you've been in this situation, you're sitting in the speaker studio and you're recording a radio play and you're this character and you're supposed to say this and then you're listening to it afterwards and you think, that sounds also fake. Let's do it another, you know, 10,000 times until it's correct.

But if we're in the real situation, just, you know, we can do it just like that. And why is this? It is because these so-called, so we have these different ways how we control our muscles and the, the two, from my perspective, most basic are the so-called volitional motor control where you have control over every little aspect. You can do spontaneous,

You know, reversal of direction, you know, whenever you decide, for example, and there's this emotional motor control and emotional motor control. Have also a super high degree of control over the movement and its dynamics probably even better, but we only have it in the, in an authentic situation. So, like I cannot.

Produce this natural prosody all the time. I actually need you, you know, to talk with so that, you know, this happens. And there's actually quite a, everyone can do it of course, but you know, like it's actually quite a high performance actually. So you think, you know, we're, using all these speech melodies and it's already going on for minutes and now there will be this full stop and here it is, you know, it's

Not only super high resolution, it's also super performance in the longer term, but, you know, we need this authentic situation for it. So I think a lot of what's happening in music is that, you know, you're playing these musical instruments, they're super difficult to play, you know, you have this extremely high note, you're Miles Davis and you're playing at this super high note.

And you're all in it for very long and you can only do it if you're really into it, you know? And everyone else will falter in between if they're not really serious about it. And all of this, you know, using music, the same is of course also true with singing, et cetera, you know? I think that's one of the reasons why also why we using singing, for example, in the first place to do something so...

Tricky with the voice is that, you know, it's a motor challenge and we can only do it really good if we're really in the emotion, in the emotional expression and we really mean it, you know? So, I mean, that may also not be true for all singers, you know, you can do a lot of emulation of emotional expression with

Voluntary and water control if you if you practice a lot. But, you know, I think that's this original idea. You know, if I mean it, you can hear it in the music. And if all this goes away from the AI generated, you know, I think people will I can I can see how this will very strongly touch upon the listener saying, hey, now

Where is all this? I mean it. You know, the music is created without this and without this, you know, honesty of signaling. So I can understand that people say, Hey, this is soulless. But of course it's a, you know, in the end,

That's, that's right. No, go, go ahead. Go ahead. Yeah. It's just really interesting because I think what that, that person, well, including myself just needs to differentiate is maybe not the audio itself that sold us because you know, audio is audio and the audio those human created did not have any soul either. It is, it is, it is your understanding of the people behind it that created it and what their intention was, what their

What the struggle was when they were making it, like that whole full emotional spectrum that you're attaching yourself to, that's the soul you're talking about. That may be missing then, you know, or you can only interpret it into the music, but the more AI-generated music we'll have, probably the more important this will become in future and the more of an extra value this will be, you know, to...

Know, this is actually, this is real. Yeah. And one of my sort of my predictions without having any research behind it is that I believe that consumers over time in general will be very aware that a lot of the things surrounding them will have, will be AI generated. And therefore there's also going to be more of a wish to attach yourself to things that has this emotional aspect to it. That I believe that consumers will be, want to be more aware to

To what is AI generated and what is not AI generated. And therefore also make more conscious choices to consume certain types of music. And also let's not forget that also this drive for AI generated music is hardly, know, the need for that is partly because it's so difficult also to have music licensed

Yeah. These, you know, like one music piece. Mean, at least that's what I heard from people, you know, could have like, you know, to even get, you know, be able to get it, you know, there's no automated fashion to say that's the music piece I want. And then you get it under, you know, there's no automated fashion of how you can get it. You know, you may have to speak with three different parties and one party is not available. And so you don't get it.

Yeah, and then it may be enormously, even if you were willing to pay for it, it may even be impossible to even get it, you know? So I think this puts a tremendous pressure also on the industry to solve all these issues that have been, you know, probably, you know,

People could be comfortable just letting them sleep for all this time. And now there's a certain pressure on the industry to say, okay, if you want your license, you get out your license music where there's real people behind and they're paid and everything. All this has to be very easy and transparent to do for users and people who want to use this music for something, et cetera. It has to be.

A feasible process. Tom, this has been a really exciting conversation. And for me, it just gives a layer of clarity to a person who spent the career understanding this topic, because there's a lot of music industry professionals speaking in between themselves of what this means. But I think at least now we have from your point, which is actually backed by years and years of careers in this.

Established some of the key differentiators between how people perceive music. I think it's an important talk in sort of understanding where this is going. And the biggest takeaway I take from this conversation is the best way that we can avoid having too big of a consequence from the edge of music is make sure we create connections. We tell strong stories that people can attach themselves to. And that's going to be our role in the music industry. Yeah.

We give people ways how to, you know, in real time control, for example, the musical sounds themselves. Yeah. Because that's, for example, the main creation that came out of my lab is we've created this technology where we give people a possibility to modulate the musical sounds themselves. So they also navigate in...

In making music spaces in a very defined way, but they have a control over it if they want to do it this way or another. And how we do it is we have them play the music through their arousal experience. So that's how we combine it with rehabilitation and with sports. We...

We have them play the music themselves through the amount of their effort. And we see that this seems to be a very efficient way how to stimulate this emotional motor control. Basically people subjectively very efficiently experience this flow state. And I think that's like dancing.

Except that you can actually change the music while you do it. And it seems that this can always, know, making the music yourself. Think, you know, that's probably the most universal way how to identify with music. Love it. Well, this has been such a great talk. Thank you for being on the Sound Conditions Podcast, Yeah, it's been a pleasure. Thank you so much for your curiosity, Jakob. It's been inspiring.

More from Sound Connections

All episodes