Ep. 28: AI and the End of Physical Sheet Music: Enote CEO Boian Videnoff

Episode Description:

Paper may have finally met its match in ⁠Boian Videnoff⁠, my guest today, the co-CEO and co-founder of ⁠Enote⁠, a music technology startup in Berlin developing powerful AI-technology capable of digitizing sheet music into fully interactive musical scores.  It’s an ambitious company with the end goal of ultimately replacing all physical sheet music.  There are some insanely cool things that the company is working on and as I told him, I am equal parts excited and terrified to see what will come of it.  Who knows where all this goes, but the potential of what might change in the next 5-10 years as the result of AI is really mind blowing and I think you will find this conversation to be extremely eye-opening.

Featured On This Episode:
Boian Videnoff

Episode Transcript:

*Episode transcripts are automatically generated and have NOT been proofread.*

Boyin Videnov, welcome to the podcast. How are you doing? Great, Garrett. My pleasure to be here.

And you’re in Berlin right now, yes? Right. Yes, I’m in Berlin. Well, I appreciate you taking the time to hop on a call with me.

I learned about your company in some press releases about, I think it was a 10 million euro funding round that you received and some other, you know, assistance that your startup had. So why don’t we just start with what, well, I should say, I was interested in talking to you both because I’m always interested in what might replace sheet music, but I’m also interested because of the artificial intelligence that your company is using to create the app, which is called E-Note. So why don’t we start there? What is the E-Note app? What is it built for? What does it do? Well, the E-Note app caters to two things.

First of all, we wanted to create one comprehensive central resource for sheet music and everything around sheet music and develop that to a platform one day where, you know, you have full access to not only sheet music, but also materials around it. So like letters of composers and all kinds of, you know, from the first autograph manuscripts to first print until then later, of course, all the publications. So really have one place where you can access all of sheet music and everything around it.

And then, of course, we wanted to create an app that gives you the possibility to use digital technology in a way that can improve your entire experience with sheet music, make it more engaging, be more interactive and ultimately replace paper. So the end game is that there is no more paper sheet music and we just use E-Note? Yes, exactly. So that would be the vision, let’s say, of the product.

The idea is that you don’t need paper anymore, that you have an app that follows your score, turns the page for you, that you can adjust the text size. You can transpose with one tap into another tonality, you know, look up from your part to the full score, have entire orchestras and opera houses fully automized in certain aspects, know that it helps children practicing and make it a more engaging experience for them. Yeah, that would be the idea.

All right. So what’s the difference between your app and other PDF viewers out there? I mean, Fourscore is one that I use. I’ve seen a couple others.

What does yours do that the others don’t? So I would say that what really sets it apart is the technology that we have up our sleeves. So we have developed the most accurate optical music recognition software that is currently available in the market. And we’re capable of recognizing symbols, printed symbols with an accuracy over 99.96% accuracy, which allows us to introduce features that other apps don’t have.

For example, we are very accurate in the counting of measures. So, for example, from a score, from a part, you can press long on a measure and it will show you the full score in that part. Or, for example, we recognize repetitions.

And then when you come to repetition, a button is shown, you press on it and it just jumps to the right place. We recognize movements, so therefore you can navigate easily from one movement to another with a tap. So this is just a few features to mention.

And, of course, all of it is also thought through very much from a musician’s perspective. So from a professional musician’s perspective. So I would argue that, you know, many of the features are placed and conceived in a way that musicians would want them to be.

And the other aspect is that we have a whole musicology team that is adding hundreds of works to the library every week. And we’re basically building a library where you don’t need to scan the music yourself and, you know, upload it to your PDF reader app. But you’d basically find it directly on eNotes.

I was playing with the app last night and there were a lot of cool things I found. Like you said, there’s a lot of the tools musicians would use, like the metronome and that sort of thing, were built into the app. So it was nice to have it all in one place.

And I’m a trombone player, so I looked up, you had the Arban Method book in there. And like you said, there were a couple of different, you know, editions of the book available. And that was kind of neat to, like, look at that and compare.

Yeah, this is the beginning, you know. This is just the beginning if you look at it. But there are, like, also hidden features that you need to maybe spend a bit more time with it.

But when you figure them out, it’s really fun because it saves you so much time. Like, for example, you can highlight symbols with our AI. For example, you can, say, make all, I don’t know, all dynamics red and then it would just turn all the symbols.

Although it’s a PDF, it would turn all the symbols then in the color you wanted or you can highlight as a single part. So there are many, many cool features that are powered by AI technology. So what features require AI and what is just good app design? So everything what is connected to recognizing something on the paper and therefore enabling the feature requires AI.

So everything that is connected to the music itself, recognizing a repetition, recognizing a movement change, recognizing any kind of symbol that you want to paint the color in a different color in order to highlight it, that requires AI. Or, for example, just counting bar lines in order to be able to count the measures and therefore being able to link an instrumental part to a full score. All of this requires AI.

Placing the metronome in the upper left corner or whatever, this is just UI UX. And also there are features, of course, which do not require AI, but I would say that just require common sense of a musician, what they’d like to have, basically. So you’ve trained the AI to understand sheet music.

Is it an image-based sort of training? Like, do you just teach it this is what the graphic looks like and it relies on the images? Or is there some other, like, is it MIDI? Or what’s the method? Yeah, it’s called computer vision. And the technology itself is optical music recognition. It’s a subset of optical character recognition, which most of us know because nowadays when you take a piece of paper that you run through the scanner and then you right away have a Word document, an editable document, which this is basically just the same technology but way more complex because in music notation you have not just few characters, but you have over 1,200 characters and over 250 of them are used in a piece regularly.

And they have connotations which are horizontal and vertical, and they have endpoints and starting points, like, for example, slurs and other notation symbols. So it’s a very complex environment with lots of crossing pixels that cross each other and belong to several symbol classes at the same time, which you don’t have with, for example, text, printed text. So this complexity creates a very, very difficult environment for AI technology to recognize those symbols.

And this is why most softwares that you know maybe that have been doing this optical music recognition usually would bring you a result which is quite bad, and you’d need to really put a lot of effort, manual effort, to correct. And sometimes it’s even faster if you just type it in from scratch. Right.

I’ve been waiting for the day where I can just scan in scores and start arranging. I’m an arranger primarily, and so I’m usually working from existing material. Is that something that the app can do or eventually will be able to do? Like, can I take a song that’s in Enote and bring it into Finale and change things? So right now it’s running just on our servers, and it’s proprietary technology that has been also receiving all this investment from the investors.

So it’s, of course, something that we try to protect because we have a competitive advantage with that in order to digitize all written music. Right now, I would say we are in the midway. So we are still not capable of bringing everything into fully digital format because we still need to train the software on certain symbol classes that have not yet been trained.

So we’re still in the development of it, but I think in the next two to five years we’ll have something that will be allowing us to have everything digital. And then, of course, we’ll try to open also the software so that users will be able to upload something in the software and get a digital file that then they can edit. So as you said, your goal is to make physical sheet music unnecessary.

Does that mean that your company is wanting to manually upload every piece of music ever written into the app? That’s an enormous job. We’re doing this since years already, yes. So we’re literally purchasing every possible edition that exists out there and we’re collaborating with publishers, we’re collaborating with libraries, with archives and we’re really trying to get in everything.

We’re quite niche already in certain areas if you search in the search of the app and you will figure out that there are already many composers that you might even not know that they existed. So is your goal to replace the publishers then? I mean, if you’re the source for all of the sheet music. No, we collaborate with publishers and I think that publishers will have the possibility to actually focus on their core business, which is discovering talent and empowering great composers to be discovered, to be heard.

I think that’s the main core business of the publishing houses and most of them already have anyway outsourced a lot of the printing, the whole logistics and even the typesetting is nowadays outsourced to companies for typesetting. So I think publishers that focus on the future and our time and the composers of our time, those will always be the ones that will have a big future. Publishers who want to earn and squeeze out the last penny of Mozart and Beethoven, they might have a problem in the future.

Are you working with living composers currently? So we’re planning to introduce a platform where also composers will be able to right away make their works available on Enote because I believe that there are so many great composers who are actually played by major orchestras but still don’t have a publishing deal and it’s such a pity that they don’t get discovered and many of my colleagues, famous conductors, agree with me that it would be wonderful and very enriching for everyone if we would be able to open a platform where a young composer can make their works available, accessible. Just imagine you write a piece for flute or trombone and piano and instantly you can inform every trombonist on the world instantly that this piece just has been published. What an amazing opportunity to get music out there and inform people about it.

Of course you need to be careful, you can’t do it with everyone, but if there are fantastic composers, young ones that deserve to be discovered, this is a fantastic way. What do you see the business model for those living composers being? Would you buy their scores to have them part of Enote then? Actually no, the business model is very simple. We distribute the revenue that is generated, we as a platform keep a part and then the rest, the majority, is redistributed to the rights owners.

So actually for the first time composers will be able to earn from it and the good thing here is that in classical music also, or in general in music, when you don’t listen to it but you practice, it’s really the time you spend practicing it. So if you spend three hours on a certain work, it will be accounted towards that work. And it’s not like when you listen on a streaming service where you just have the click or the first 30 seconds accounted.

Now that’s really fascinating. So you’re saying that musicians’ practice time would also be factored into what the composer would earn as like a percentage of revenue? Of course. That’s really, really – okay, you’ve blown my mind a little bit.

So do you – I have to admit I’m a little bit torn because on the one hand this all seems super cool. On the other hand, I do kind of worry about if all music is one – I mean this is the Spotify thing, right? Like nobody makes any money off Spotify because there’s so much music the payouts end up being teeny tiny. The problem in Spotify, if I may say, is that you have one industry that is dominating majorly the audio streaming, which is today’s basically entertaining music, pop, and other genres.

And who is not really making money are, of course, all the niche and classical. Unfortunately, we basically don’t make any money anymore in classical with audio streaming. It’s all just promotion.

But also the model of if they would actually have the usage counted and not just – you know, if you listen to an opera, it’s not the same as if you’d listen to a Taylor Swift song. It’s just the first 30 seconds that count. And the opera, sure, we can divide it in many, many tracks, but still it’s not the same.

Therefore, the listening time, there are some people who listen only to certain genres. And if this would change – I’ve been advocating a lot also for a user-centric model, which I hear already that the major publishers are considering, which is a fantastic thing because I know that Spotify has been also pressuring in that direction. User-centric means that you really look at the usage of one specific user and distribute the money that this specific user has paid for the service to the right holders that he has been consuming.

I find that very fair because then my 10 euros get distributed only to the ones that I have been using, and they’re not being put in one major bucket and then distributed accordingly to the region of all users. How is that even possible? Is that what the AI does? Or is it just the fact that you’ve built the app that way? Right now, we are still in conversations with the major publishers. Our major innovation is that we have this – or not innovation, but let’s say our major difference from the current audio streaming is that we do this seconds-based, so the seconds that you have spent in a score are accounted for, so it’s time-based.

But my ultimate goal and wish would be that also music streaming one day not only does it on time-based, but also on user level, so that really your payment gets distributed to the right holders that you have been using and my payment to the ones that I’ve been using and not that our payments get put together and then we account for the total time that you and me have been using it and therefore distribute our money together to the right owners. Do you understand the problem there? Because you have basically one person who might use it much more, distributes the money of the other who uses it maybe a bit less. Yeah, I think that’s a really fascinating way of looking at it.

But I also wonder, do you think that will ultimately end up with less money going to composers? I mean, because if I’m paying my €10 and I’m listening to music all day, every day, that’s not going to be very much money for the artist that I’m, you know, if I’m a performer and I’m performing all these different works, I mean, they’re going to be getting less money than potentially if it came from the larger pool. I mean, composers nowadays have two major ways to earn, right? They have the commissions and commissions, I think, is something really important that needs to stay. When there’s a commission, a composer makes a living with this.

And of course, from the rights and the rights when works are performed and those rights are not touched. So if there is a performance and a composer is being played, those rights are still being paid by the promoters and this is how the composers get. So this is something that they would receive on top.

Basically, they’d get money for the practice time of musicians that currently they’re not getting any money for. Yeah. I’m just so fascinated by all of this and I really think it’s compelling.

I just wonder, is it realistic? I mean, even just to get people past paper. I mean, the e-book industry has been trying to do this for a long time, right? And we haven’t seen a shift in the same way. I mean, most people are still saying that they prefer a physical book to a Kindle or some other sort of device.

So how do you think musicians can get over that hurdle? How do you think you can convince them to give up their paper? I think it’s actually a very normal and natural thing. There is a change of habit that is happening. We’re seeing also that there are certain regions in the world that are much more digitally advanced and ready to do that.

For example, we see users in the US and in South Korea, for example, who are very, very digital. I know basically almost all my friends who are US musicians have an iPad and play somehow also from an iPad. So there’s a cultural thing also happening.

Here in Europe, we are usually a bit more behind in those kind of things, but also because the device itself that we currently have, the devices are not perfectly suitable for a professional institutional usage. So I truly believe that there needs to be a device that will be specifically tailored for a professional usage, especially within orchestras. We’re actually talking with one of the major hardware producers in the Silicon Valley to have a joint venture about that for the future.

And then there are iPads and tablets that are great for if you do it as a hobby, you have then access to everything. You can make it an interactive experience. I really believe it depends on the features and how much people realize that they make their lives so much nicer and easier.

When people realize that something is so much better, then they change their habits. It’s very simple. If you would go today to a lawyer’s office and ask them, do you guys want to work again with typewriters or do you stick with Word? Nobody would tell you, oh, yeah, sure.

Typewriters were so nostalgic, that noise in the office that was so nice. I mean, this is really, really simple. Bring innovation that is meaningful, and then people are going to adopt it.

So you’re also developing a hardware device that’s just for music? Is that what you’re saying? Or hoping to? We are exploring that option, yes. And I can say only that we are very excited about that. And I think that that will be the major game changer for the professional institutional usage.

Yeah, I mean, you make a good argument. I think in theory, right, like if everybody just agrees and adopts this, it fixes a lot of problems, like you say, with the payments and the practice time. And it would also solve a lot of the issues about copying and illegal downloading and that sort of thing that you see in sheet music.

But, yeah, I just feel like musicians are stubborn. I’m not so sure that I’m convinced they’ll all go for it, you know? You know, it takes time. It’s nothing that happens overnight, especially not change.

I remember that when Tesla was founded, people were laughing at Elon Musk that there would be electric cars. And now every major car producer has pledged to stop producing combustion engine cars in the future. So it really needs time.

And it’s also fine because, you know, to build a perfect product that really would replace paper needs also time. Paper is a pretty good product. It’s a great product, but it also has lots of limitations.

I mean, if you want to change a misprint, you can only draw on top of it. You just cannot grab the note and move it a bit down and then just have it fixed. You know, if you want to make text bigger because it’s just too small for your eyes, what do you do? You copy it and print it bigger.

Well, why don’t you change the text size, like in a Word document? So there are many, many arguments also that speak for a digital solution. It just needs time to build the perfect one, and it takes a long breath, and it takes, you know, the beginning always requires people who believe in the vision and join early on, and then at some point, you reach a point where you convince most people, and then most people join too. So how many years do you think it will take for this to really scale and come into its full potential? I think that the next five years will be crucial in our development.

The five years from now on, our goal is to place E-note in the educational market and make basically it a standard solution for music education. Also so that, you know, kids who grow up with it, it’s something natural for them. You know, we have a digital native generation that we’re teaching still with tools from the 19th century.

I mean, it’s not that engaging as a video game or any kind of interactive experience that comes along their way, and we’re seeing that motivation also to practice is not at the highest levels among kids. Also because they’re left alone. You know, you have one lesson, 45 minutes to one hour per week, and then you’re left alone, and you should practice.

If you can guide that practice in an interactive way where you can make it like a bit gamified and, you know, at the same time teach actually methods on how to learn, then you can build this passion of music making in the children. And music making is proven that, you know, American Association for Psychology has actually made a big study about it, and they have found that every child who plays actively an instrument is one entire academic year ahead of children who do not play an instrument. So it’s incredible, the effect.

And yeah, this is what we would like to do. Make it more engaging so that we can bring music making back to the center of, you know, every people’s life. What are the copyright implications of all of this? I mean, if composers are putting their music onto the app and it’s possible to, you know, like you said, to change notes, to change keys, to alter things, does that create problems down the line? No, because it’s your personal copy, right? So if you make a change… Actually, I think there is even a community benefit because if 10 people make the same change in a score, most likely it’s a misprint.

And we actually could use that community effort to find misprints and correct them, the scores. But if you make your own changes and no one else can see them, that’s your copy. I think that’s fine.

So how have your conversations gone with publishers, if you don’t mind me asking? Because I see… Well, you know, I just feel like the sheet music industry in general is quite behind the rest of the world in terms of technology, right? I mean, you have all of these AI startups and things happening on the one hand. And then over here in the sheet music industry, we’re arguing about whether or not it’s okay to preview the whole PDF before you buy it, you know? And it just seems like such a big leap from where we are now to the world you’re talking about. I mean, PDF is a dangerous medium in that field because musicians, when they get their hands on PDFs, they’re never going to buy anymore the publication, right? So I totally understand that publishers are worried about PDFs.

And PDFs is also not the right medium for digital right management. So when you have work that has been written by someone and that is rightfully still protected, then this has to be protected also in a digital world with all possibilities that we have. So to make it short, the publishers who have copyrights and who have works and composers that have copyrighted works, they’re very open to work with us.

We’re signing deals actually currently right now with most publishers, and we are in conversations with the others with whom we are not signing, and I expect this also to happen. So I see that there is a big openness. At the same time, of course, I wish to bring on the points that are mattering to them, that matter to them and are important for them, and we’re open to hear those and find a future that is viable for everyone.

So if you had to pick, what are the top three features that Enote offers? What are the top three things that would convince somebody to download the app and start using it right now? I think the score viewer in itself, all the features that it has, is the most thought-through score viewer currently available on the market. So you will find that it offers so much more, so more intuitive and easy to navigate. So this is something that, because music making is a complex thing, it’s now, if I tell you, yeah, you can navigate from one movement to the other, you can jump from one repetition to the other.

Sure, I can go in this granular level, but I think what really makes Enote strong is the score viewer in its entire conception of how it works. It is just very convincing. And the other feature that I love personally is the discovery possibility in the search.

So when you go in the search and you can find pieces for violin and piano from a French composer who has anniversary in 2030, and it should not be longer than 10 minutes. You can make a search like that and it will filter the entire catalog. And imagine one day all of written music, it can just filter it and you can find works like this, where now you need to call a friend maybe and say, do you know something that could fit that? So that’s really something that I adore.

Is there audio playback in the app? We have a MIDI playback that we will be doing for our previous course that are fully digital. Actually, we’re working on a natural audio synthesis based on a project Google had been starting, where they have built an incredible AI algorithm that is capable of playing back MIDI files with a very, very accurate, realistic sound of a piano. Very interesting project.

So we’re developing that further. And we’re bringing actually pretty soon YouTube recordings and integrations to streaming services. We’re also partnering with Deutsche Damophon, for example, where we’re bringing their catalog on.

So yeah, we’re bringing music so that you can listen to the score, listen to the music and follow the score. So the end goal really is to make this more than just a sheet music app? Yes. Because you’re talking about listening to all the different versions of the piece, as well as looking at all of the different editions of the piece, but then also being able to, I guess, use AI to mark it up in a way that’s more efficient? Yes, yes.

This we can do already now with the AI. It saves me so much time. I find myself being able to study in places where I was not before.

I mean, I’m a conductor. I need usually a ruler. I have my pencils with different colors.

I need space to open the score with a ruler, make my markings. I now find it so easy because making structure, for example, I just tap on the bar line and the bar line gets highlighted. And things like that, I can work in a subway or in a train where before was totally unthinkable.

It saves me lots of time and makes it more convenient. But yeah, our goal is to make it a platform. So really the idea is to evolve it and make a platform where students can connect with teachers, where music lovers discover new artists.

And 10% of all streamers play actively an instrument. That’s over 200 million people who play actively an instrument every week. Imagine if you’d be able to, when an artist brings out a new album with AI right away, create new arrangements for different skill levels, for different instruments.

Imagine that possibility of generating playable scores for everyone, for the real beginner to someone who’s already intermediate. How much more you’d bring music making, active music making back to society. I think that’s super exciting.

How far away are we from artificial intelligence that can, as you say, listen to something and create the score? Not far. I’d say a few years. So are we also going to have, I mean, there’s already AI that can compose songs, right? Are we soon going to have AI that can also compose and then notate it? Yes, absolutely.

So I think we’ll reach a point where generative AI will not only be able to generate audio, but will be able to generate actually the sheet music of that audio and even on requests in a specific style or skill level. So what would you say to the composers that are freaking out right now, thinking that in five, ten years, they’re going to be competing with all these robots that can be composing 24 hours a day and churning out music? You always need to see for what use case, and how many composers do compose for a kid who’d like to play an Adele song? I don’t know any composer. Well, there are people whose whole job is to transcribe the Adele and to make educational arrangements of popular songs.

I mean, that’s what a lot of, especially for piano and guitar and those sorts of instruments, there’s a lot of lessons based around popular music. Of course, but the skill level of that specific child is not always matched. And so you have either very, very simple arrangements or arrangements that are not playable for the current state.

So what I’m trying to say is that when you employ technology for an application case that you actually cannot handle manually, it’s useful. Will it have huge artistic value? No, I don’t think so. Because ultimately, we humans are interested in the story that is told in music, and we’re interested in human stories.

So I think what the composer puts on paper or what an artist writes when they have a new song, this is what we’re interested. Not in the arranged version now for that specific use case that the child would like to play it. Or maybe if you want to have a jazz standard and a certain style arranged for you that you’d like to play, that just actually enhances the amount of people who then can play music.

And therefore, it enhances empathy in our society, makes our society a more creative place. And ultimately, that’s what we need in an AI-driven age. Because AI is taking over most of our professions over time.

And if we are not becoming a more creative society and a society that embraces individuality and creativity and diversity and also trains that in us, then we will have a problem. So I think we need to bring music-making back to the center of society in order, actually, to find our new role in this new AI-driven age. Okay, putting aside my panic for a minute, how is it possible for AI to evaluate the skill level of a player in order to create something for them? Yeah, that requires training, of course.

So there are a lot of pieces that have been already rated in skill levels. And that, of course, helps through comparison, then, to rate also other pieces or basically create sheet music in a certain skill level. Let’s say if you have… Every time you see octaves in piano, let’s say you know that skill level 4, but in skill level 3, there are never octaves, then the AI would know that for skill level 3, should not put octaves, you know? Things like that.

Very simple and maybe oversimple example, but just to make the point. So it’s always based on lots of training data. So the more training data you give, the more accurate the AI is.

And based on the training data, then it can generate or it will be able to generate arrangements that fit that criteria from the data it learned from. How does this kind of AI compare to, like, chat GPT and the kinds of AI that are commonly talked about in the news? It’s very similar. It’s the same kind of generative AI.

And that’s also based on, you know, understanding language. Basically, ultimately understanding, right? When you talk to chat GPT or when you chat with chat GPT, you have the feeling you’re speaking to someone who understands what you’re asking. And the same thing is also then with music.

It’s a language. So if somebody understands the language of music, then they are also able to, you know, articulate themselves in this language, which means compose. So what do you think, I know this is a hard question to answer, but what do you think then the future of the music industry is going to be, you know, 10 years from now? Because this is going to disrupt a lot of things.

I mean, you’ve got traditional publishing, you’ve got composing, you’ve got songwriting, you’ve got all these different, you know, I can see good and bad things happening as a result, you know, depending on how you’re currently involved in music, right? I mean, if your whole job is transcribing audio that’s been performed and notating in that, like AI is going to put you out of a job. But if you’re somebody that’s a student and performing, like this is incredible. Like you have so many tools now to help you.

I mean, as with any technology, there’s, you know, I don’t want to call it winners and losers, but you know, like it’s going to change things for better or for worse. What do you see the biggest changes being? Either you embrace progress or you fall behind. I don’t think it makes sense to fight progress because it happens.

Whether it’s Enote or another company, this will happen. Now the question is, what do we do with it? And you’re totally right. There will be professions that will change as they’re now changing.

I mean, I was in San Francisco in a driverless taxi. That’s happening too, whether we want it or not. And we need to understand, and this is what I was talking before about, that we as humans have a new role in this new age.

Our whole educational system is trimmed to a industrial age, where we educate a uniformly shaped human workforce. And all of this is challenged by this age. We need creative people, empathetic people, out-of-the-box thinking people.

And those kind of qualities will bring us to find our new roles. And yes, if you are in a profession that can be replaced by a machine, then it means that the education was wrong. We need to change our educational system.

We need to rethink it. Is it really necessary to teach things that a computer can do and everyone has in their hands with their phone? Or maybe we should teach other kinds of things that will give us the possibility to broaden our horizon and go into professions where creativity is necessary and where machines cannot excel as well as we do. So I think it’s a bigger question and a bigger problem that you’re addressing.

It’s not just music. And music, as every other field in life, has a part in it that consists of more technical, let’s say, let’s call them more technical tasks. And there are also more creative tasks.

And I think the same as in other fields, where we see machines take over the more technical parts or the more repetitive things, it will also happen in music. And the more important it is to shape our creativity and find people who can voice their stories and make them engaging and amazing so that people love it and want to hear more of it. So how can musicians who don’t really understand the technology prepare for what’s coming? Because I don’t really know much about AI.

I’ve been trained as a musician, right? I think great technology is great when you don’t notice that it’s there. So you shouldn’t be caring about what and how it works, that when you reach a point that there is a repetition, there is a button that is shown. You just should be caring about pressing that button so that it brings you back to the right place.

It should not be something that you now start to think, oh, how to understand this. So sure, there is a learning curve once you have a lot of possibilities that you couldn’t think of, and you need to discover the features, how to use them. But this is more of a first-time discovery.

But the moment you start to think about the technology and what it can do and how to do it, I think that’s not the right approach. We don’t need that as musicians. We just need helpful software that makes our lives easier, more convenient, and more engaging.

And if the technology is done well, then this will happen. How can the app help composers that are creating their own music? Are there tools for that? For now, not, but I have this, you know, this wish. Imagine you could, I mean, when you have all of sheet music at your disposition to be analyzed, you basically can let big data analysis run over it.

And besides that, you can discover maybe many things that haven’t been discovered until now, like citations and, you know, things. But imagine then if you tell a young composer, please compose in the style of something, you’d be able, based on data, to tell them, look, this kind of progression, they never did. Like, this kind of progression doesn’t make sense, but it makes sense like this.

So you could run analysis and give them feedback on something that they’d need maybe otherwise to have access to a great teacher. Not everyone has access to a great teacher. Also, I think the same for, you know, so many children would love to study with a great artist, with a great teacher, but not everyone can.

There are a few spots. So if we could get the great pedagogues to, you know, bring in their knowledge into such a software, then we could share the knowledge with much more, many more people out there. So I think that is really exciting.

My head is sort of spinning right now. I mean, it’s really wild to think about and fascinating to think about, you know, all of the possibilities that, you know, are on the way. Why don’t we end with the story of how you came to be involved in all of this? How did you start the company? How did you start the app? I mean, what was it like getting all of that off the ground? I mean, when I started this together with my colleague, actually it started with two of us, Joseph and I. Joseph came from IT management background and I was dragging him to my concerts.

We were friends already in university and then stayed friends and became now business partners. It all was basically an idea of bringing all of sheet music into one place and making it more engaging because I grew up in a, let’s say, a family that liked digital gadgets. My dad always had a passion for it.

So he passed it over to me. I always liked the newest trends in technology to follow what’s happening. And it felt always that there is a big disconnect in our industry.

And it felt also that we could actually profit a lot by the technology that is out there if used correctly. And then, of course, this idea grew and grew in conversations and how that could be actually applied in the music education and how music education would benefit and ultimately society would benefit. So, of course, such things grow, right? In the beginning, you start by wanting to solve a problem that you have and then you end up seeing how many more fields this touches and how big this actually can become.

It’s not an easy journey to found a company. We are now over 60 people working on this. It’s a lot of know-how that has been built.

We have specialists from many different fields, from musicology, from development, that now speak the same language. It took years to build that. And, of course, also lots of R&D, until now €20 million that went into all of this.

So it’s an expensive endeavor, but it’s one that is really worth it. And all the sleepless nights somehow feel that are still worth it, yes. Do you think there are any limits that need to be placed on this kind of technology? Well, artificial intelligence is a dangerous thing.

I agree with that saying. The question is, where do we apply it? So I think the limits, as long as we stay in charge, we humans, we’re putting limits. When we’re not anymore in charge and when it’s being misused, and unfortunately all kinds of technology can be misused, and sometimes also not really on purpose, but just because it’s there.

So, yes, I think we need to be very aware for what we use the technology for and in which cases maybe it shouldn’t be used, but we’re still talking about music in this case. Right, right. Is there also going to be any way of knowing when music was created by an AI? Thinking about copyright and thinking about also just, if you’re only listening to it, there’s certain types of music where it’d be difficult to tell.

Yeah, I mean, this is already now a topic, right? When you spin off something or when you create a song in the style of an artist with his voice through an AI, now who wrote the song? Who does get the copyright? So these are all topics actually that are currently being discussed and I’m not 100% on what the current developments are, but I know that they’re part of the active conversation and worries. So, yes, this needs to be discussed. This needs to be… Usually legislation comes always way after the innovation happened.

We adapt always way too late to it, and I expect that there will be many cases where people will be using something and then later, many years later, we will have also a law for it. Well, thank you again for taking the time to talk to me. This is equal parts fascinating and terrifying and exciting.

Is there anything else you wanted to share with us before we go? No, I’m very happy that you reached out. It’s always a pleasure to talk about this passion of mine and don’t be scared of it. I really think embracing the future and trying to shape it is less scary than just watching it happen.

So this is maybe one of the reasons why I got involved because I prefer to be actively shaping it rather than standing by and watching it happen. Well, thank you again. I really appreciate it.

Thank you so much, Gareth. It was a pleasure.