Malte Giesen

Malte Giesen–in conversation with Alexey Munipov

 

You are both an active composer and the head of the studio for electroacoustic music here at the Academia Kunst in Berlin. How do you divide these roles? Are they completely separate in your life, or do they inform each other?

 

In principle, they are different roles. As a studio director, I have a very different function than I do as a composer. I’m more of an enabler for projects of very different kinds, from very different disciplines–and that requires me to step back from my own aesthetic and artistic preferences and take a more overall perspective. In that sense, I switch roles: I’m more of a servant to other artists. I have to grasp their vision and help them realize their work.

 

As a composer, I’m much more »artistic« in the sense that I’m focused on my own work. It’s a narrower, but very concentrated way of looking at things–and it allows me to go deep into detail and explore things very thoroughly. I can’t always do that as a studio director, because there are so many things going on: helping people, setting up recordings, taking care of the technical side, coordinating with other departments of the academy. I try to keep a balance, but at the moment, the academy work takes up much more space than composing.

 

Do you have an internal rule–composer first, director second, or vice versa?

 

Not really. There are other roles in my life as well: I’m a father of two children, a husband–there are other duties. But when it comes to working on pieces, I try to keep these two roles separate. Of course, they benefit from each other. As a studio director, I get a lot of ideas from the projects I’m taking care of. And the other way around, composing and exploring technology gives me the opportunity to develop the profile of the studio further. So it’s a give and take.

 

The studio has a unique history and a strong identity, with a lot of singular equipment around us. Do you use that for your own work in any way?

 

The historic instruments–not so much, actually. When I started here, I barely had a connection to them. Over time, I became more and more aware of the special features–and of the treasures we have in the studio. Working with the old instruments, and talking to people who were active during the period when these instruments were developed, changed my perspective and, I think, enriched it.

 

Today I have a different view of electronic instruments–and of what an electronic instrument is in general–than I had before. It’s not simply obsolete technology. Instruments from a certain period can do certain things and cannot do other things–and that is actually good for creativity. You don’t want a universal machine.

 

That’s something I discussed with Robert Henke, one of the co-founders of Ableton. He always said that people don’t want machines that can do everything; they want tools that do certain things exceptionally well, and other things less well. Exploring the boundaries and limits of machines–including what’s glitchy, strange, or imperfect–is where it becomes interesting.

 

So what is the strong side of these older electronic machines, in your mind?

 

I would say: directness.

 

In Silicon Valley they’re now talking about calm technology–about getting away from screens. And the most obvious advantage, to me, is that you don’t have a screen. You have hardware: knobs and faders, cables you pull, plugs you connect. It’s a more haptic, physical way of interacting with an electronic instrument–and of creating sound–than typing on a keyboard or clicking with a mouse.

 

And it produces different results. It’s also more compelling in a performance situation. Watching someone turn knobs and push faders is simply more engaging than seeing someone behind a screen, where you can’t tell whether they’re checking Facebook or emails, or coding sound.

 

I know at least one work of yours that uses these machines–the on with the white noise.

 

That white noise comes from the synthesizer over here: the only analog modular system that was built in the GDR. It’s from the 80s–but even then it was already old technology, because they didn’t have access to the most recent developments like they did in the West.

 

All of the sound-producing modules–oscillators, filters, the noise module–have a distinct color. White noise is a good example: you can generate it mathematically and it becomes super pure, super clear. But I wanted an analog white noise, with impurities–with the natural fluctuations that come with it.

 

So it’s a kind of luxury vintage white noise.

 

It’s a little like luxury watches: they are mechanical. Of course, quartz watches, electronic watches, smartwatches–they show the time much more precisely. But it’s what you said: it’s about the emotion the object conveys. You can buy a mechanical luxury watch for tens of thousands of euros, and at the end of the day it’s still »just a watch«–but it has a different aura than, say, an Apple Watch.

 

But sound-wise, it did sound different.

 

Of course. It sounds different because it’s analog and has its own coloring. The digitization process also plays a role: every digitization of an analog sound produces artifacts, depending on how you digitize it. But yes–it’s audible. I think you can hear that it’s analog white noise.

 

Could you say that the studio itself has its own sonorous quality–like it’s an instrument, with a timbre of its own?

 

Absolutely. The studio is an instrument. It’s the selection of components, the tools and setups you have–and also the people working there, the way knowledge is shared. The social component matters too: the exchange between people in the studio produces a certain kind of sound, or at least a certain range of sounds. In that sense, each studio becomes a unique instrument in itself.

 

Have you ever tried to compose with this instrument–with this very studio–as a composer?

 

In a way, yes. My NFTs pieces were a discursive work around this environment–around the studio and my thinking about analog electronic instruments at that time. And more broadly, it influences everything I do. Everything I experience here, everything I produce here, everything I create here inevitably feeds into my work as a composer.

 

And more broadly: does it matter to you that listeners can recognize your music by ear–that you have a personal artistic signature, a style?

 

I think there’s a gap between how composers imagine this and how listeners perceive it. Many composers feel their pieces sound very different from one to the next–and some will say they have completely changed their way of composing, and the way it sounds. But listeners often still say: I recognize your style very clearly, even across different pieces.

 

I’ve never felt the need to push that. I don’t try to »take care« of having a signature sound. I think it happens anyway. That’s also what I tell students: don’t worry about it. If you work steadily, if you keep composing and producing music, a certain style develops on its own. It will be there no matter what you do.

 

I also had the impression that you care more about concepts than about sensorial qualities–about sound itself. Is that true?

 

Yes–at least in how I start. The conceptual side is usually more important to me at the beginning, because everything derives from it. And of course, that’s personal: other composers work very differently. I know people who work more bottom-up–they start with one sound, and then it grows, and they develop deviations and variations from there. That’s mostly not my approach.

 

For me, it makes sense to first clarify what the piece is about–what the work is about, what the core conceptual idea is. Once I have that, it becomes very easy to derive everything from that core.

 

Does it matter how it sounds in the end?

 

Of course, it matters. But the way it sounds has to be justified by the concept. It cannot be arbitrary. The sounding result is important, but any sound can have a lot of artistic force when it is strictly tied to a concept. It’s about contextualization: you can take almost any sound and place it inside a strong aesthetic and artistic idea.

 

Let’s move to your new piece–we don’t know much about it yet. The programme mentions something like an AI-controlled diffusion of white noise. Could you unpack that?

 

Over the last couple of years I’ve been working with generative AI, and I dug deeper into how these models work–the processes, the mathematics behind them. What struck me was that generative audio works in a very similar way to generative image.

 

One of the most famous algorithms for generative images is Stable Diffusion. To put it simply: you train a model by feeding it pictures, and during training the model adds white noise to the images iteratively–step by step, more and more noise. At a certain point, you no longer recognize the image because there is too much noise.

 

The model then learns to reverse that process. To be clear, there are intermediate steps–there’s also the latent space–so it’s not literally just »the picture,« but let’s keep it simple for now. When you prompt the model–say, you ask for a horse on a skateboard–it starts with noise and then subtracts noise iteratively, until only the horse and the skateboard remain.

 

With audio, it’s basically the same–except you don’t work directly with audio files, not so much with time-based media, but with spectrograms. You convert audio into spectrograms–pictures again–and you train the algorithm on those. Then you can generate new spectrograms from prompts, synthesize them, and you get sound. That’s essentially the process.

 

What I’m trying to do is recompose that process–to find an acoustic or musical analogy to it. So I take small snippets of a composed structure, and then I noise or denoise them in different steps.

 

And this doesn’t happen only by adding white noise on the surface of the sound. It also happens in the so-called latent space. The latent space is a representation of these short snippets, organized by proximities of different sound properties. A piano, for example, is »closer« to a marimba than to a string sound; a string sound is closer to another sustained instrument than to a plucked one.

 

This applies across many parameters: rhythm, harmony, pitch, and so on. I try to run these noising and denoising processes on multiple layers of that latent space. So it’s not just »more white noise« or »less white noise« overall. Rhythm can become more noisy–or more denoised. Pitches can become more unstable and noisy–or denoised. You get different degrees of recognizability of the musical material.

 

You can also morph different musical materials into each other. One part is like a quote from another piece I wrote several years ago. Then there are four bars of easy-listening jazz. These different snippets are morphed into each other, again on different levels.

 

And there is not only music–there’s also a video part. Because this model is based on generative image and generative video, I wanted an additional »player« within the ensemble: a vertical screen–like a smartphone. A big smartphone on stage, as an extra performer. It doesn’t run all the time, only at certain moments. There are snippets of generative video with different structural purposes, drawn from different sources.

 

One of the most famous examples of generative video–something that went viral–are these clips of fruits or glass being cut by a knife. They look super smooth: super clean, super nice-looking glass flutes, and they can be cut, which is physically impossible. They have that ASMR feeling, which has gained a lot of popularity in the last couple of years.

 

I took this idea of cutting glass without really having a source, and transferred it to the instruments. I generated videos with the instruments the players are supposed to play, and tried to recreate these »cutting« videos with instruments–basically to see what happens: how I can prompt the model, and what kinds of things occur when I do that.

 

It also refers a little to the idea of instrument destruction–mostly known through that famous performance with the violin. I recreated that, in a way, through generative video. I think it’s an interesting new technology. We’re at a moment when this technology is still at the beginning. You can play around a lot and get results you couldn’t achieve before.

 

So it’s AI-generated white-noise-derived meme-induced piece of music

 

Yes. That’s a good reduction. But it’s still manually composed. I would say: manually composed, but inspired by AI algorithms.

 

What is the ensemble’s role in that?

 

My main interest is still the live situation: people on stage making music together. Most of the instruments are analog–with the exception of the keyboard, the electric guitar, and the virtual drum set instead of a real drum set, for the reason that I want to switch through different drum sounds very fast.

 

Like a generative algorithm would. When you say »horse« to the algorithm, it has thousands, ten thousand, millions of representations of horses. And that’s the same thing I wanted with sound–for example, with drum sounds. I wanted access to a couple of hundred different drum sounds, quickly.

 

Is there any message behind this concept?

 

Not a message–not a political message in that sense.

 

I think it’s mostly about playfulness: discovering things, playing with tools, and finding the limits of those tools. Curiosity, playfulness–that’s the main motive. These are new technologies. A lot of people still don’t know how to use ChatGPT, for example.

 

It’s super recent, and it will change a lot in the next couple of years. But for now, it’s something you can play around with. You don’t have to know every technical detail perfectly. And that can be fruitful: to explore the limits, and to let the audience see your process–your interaction with the technology, and the dialogue you have with it.

 

Let’s discuss AI for a moment. AI is a loaded topic. In the industry, a lot of musicians react negatively–and the conversation often turns quickly into fear: »it will take everyone’s jobs.« Meanwhile, truly convincing artistic uses still feel relatively rare. How do you personally relate to AI as a tool for making music today?

 

I would say there are two answers to that.

 

The artistic, idealistic answer would be: in my observation, people are not interested in AI products–and not interested in AI music–because music is a human praxis, not a product. People will continue to make music out of interest, often out of curiosity. People love to play instruments themselves.

 

The fear of being replaced comes up when music is treated as a product with an economic value: production music, music for movie trailers, generic films–situations where it’s cheaper to prompt an AI, and mediocre quality is sufficient. In those areas, musicians will be replaced, and composers will be replaced very fast. And I actually think that’s a good thing. You don’t have to do all this generic, cheap stuff.

 

The stronger impulse, then, is to make something that stands out and is more original. Maybe generative AI will push people toward being outstanding–toward being more experimental. And maybe it will give a push to new music, actually.

 

I think in the end people lose interest when things are obviously AI–or when they find out everything is AI. I saw this when Coca-Cola launched its Christmas advertising video a couple of weeks ago. They communicated that it was completely AI-generated–and also super expensive, because they had to use something like 70 million prompts to get the whole video.

 

It cost about the same as if they had produced it traditionally, because they used so much compute. And people hated it. If you read the YouTube comments, everybody hates it. It’s super high quality, but you still see that it’s AI, that it’s generative video–and it’s a little bit worse.

 

It’s like someone giving you a birthday cake: they didn’t make it themselves, they bought it for $0.99 at the supermarket. You think, okay–they didn’t put much effort into it. I think the evidence baked into the work plays a big role.

 

Fifty years ago, electronic music was full of optimism–it was the cutting-edge frontier. There was this belief that you could generate any sound and maybe even do without musicians, and many composers found that idea exciting. Now it feels like this has actually become possible–and yet it doesn’t make people happy. The industry mood is much more pessimistic. Why did the electronic-music dream of liberation turn into today’s fear of replacement?

 

Maybe because we are liberating people from the wrong kind of work. We don’t want to be free of making music, or drawing, or writing books. We would rather be free from housework–cleaning up. And now AI is taking over the fun stuff.

 

And when I read older texts–from the 50s or 60s–by Stockhausen or Hans Eisler, where they say, okay, in the long term musicians are going to be replaced by electronic sounds and computer music, I always think: wow. When you read that today, you see how naive it was, in a way.

 

So when I look at today, I’m not that worried. And when I look at aesthetic developments in the last couple of years–especially in the studio environment–it doesn’t follow the rules of technological development. It doesn’t follow the rules of economy, or efficiency. People are anachronistic. Electronic music people are moving away from screens and going back to hardware.

 

There was a huge boom of analog synthesizers, modular synthesizers. People love hardware instruments. People still work with tape, or return to tape and analog formats. Last year, overall revenue from vinyl surpassed the revenue from CDs. And when people publish things, they publish them on vinyl–not on CDs–and they put them on Bandcamp. So for artistic work there’s a different logic that doesn’t fit market logic or efficiency logic.

 

 

And there was this idea: electronic music is the future, it’s cutting edge. Even the name of the laboratory–»laboratory for acoustical musical problems«–sounds brilliant. Now it’s not about grand problems anymore. It’s about: let’s take a look at this nice retro, soft sound of the past.

 

And in a way we live inside that old visionary dream: everything is electronic. Even if you record a string quartet, it’s tweaked and edited, you use effects. Everything is electronic.

 

For decades, electronic music was framed as »the future«–cutting-edge, even utopian. Today the discourse often feels more nostalgic, drawn to retro warmth and older tools. We’re living inside the old dream in a strange way: almost everything is electronic now–even a string quartet recording is edited, processed, mediated.

 

For most people today, the listening situation is 99% headphones, not acoustic instruments. So the analog instrument is the exception.

 

And yes, it’s exactly like you said: electronic music is not about cutting-edge technology, not about being the technological avant-garde. Today, I feel it’s much more about a kind of mindfulness toward technology–toward all technology–recognizing the unique qualities of imperfections.

 

All the technologies you have–synthesizers from the 60s, laptops, digital effect devices from the 80s–they are on the same level. The old synth is not inferior to the laptop; it just has its own qualities, like the laptop does. You can use all of these technologies from different decades within the same work. You can use whatever you need for your artistic vision and other purposes.

 

Electronic music has this constant anxiety: everything becomes obsolete. How do you deal with that–the sense that what you do might be devalued quickly just because the technology moves on? For example, a few years ago you made a piece called Sound for Rich People–an NFT piece, at a moment when NFTs were the hot topic. Today, you need to remind people what NFTs were.

 

In terms of technology, that’s actually something we discuss a lot at the academy–especially in relation to the archive and the question of keeping works alive, particularly works with live electronics. We receive archives from members of the academy–for example, Thomas Kessler, who died last year–and in his archive there are many works with live electronics: patches programmed in software that is 25 or 30 years old, which nobody has anymore.

 

We’ve put a lot of thought into how to preserve this kind of work. Preserving sheet music is easy: you put it in a box, and that’s it. With electronics, it’s different. We’ve been thinking a lot about how to preserve electronics–and actually large language models can play a big role, because they can convert one programming language into another quite easily. They »know« the details. That can be one way of preservation: to keep works performable, to keep them performative.

 

Because it’s a living praxis. You also have to think, even today, about how you perform a piece by Bach: you don’t have the instruments from the period–or you can only find a few–and they were played differently. So it’s always an interpretation of the work. And when you realize or recreate electronics, it’s also interpretation. If you take a piece by Luigi Nono–you have to recreate things digitally. You don’t have the original devices. You have to patch it in Max/MSP. That is an interpretive process as well.

 

As for relevance–as for society’s attention–I don’t think you can plan for that. You have to do what interests you, what fascinates you. Often, it becomes irrelevant a few years later. Sometimes it takes fifty years and then it becomes relevant again. But you can’t plan it. I don’t think it plays a role in my composing.

 

And speaking of generative music: it’s not a new idea. People like to trace it back at least to Mozart’s dice games–and in the 20th century it becomes a constant dream: generating music instead of writing it note by note. Now we’re at a point where it’s routine. How should–or could–composers use this technique today?

 

I can only say what I find interesting from my perspective, at the moment. Right now it’s more of a tool than a replacement–and it’s very good for sketching. If you want to try things out quickly, and get a lot of variance, a lot of different results, just to test possibilities–you can do that extremely fast now. But if you want to control details, you still have to do the manual work. Fine control is still very difficult.

 

For example, with so-called »pseudo-audio,« you can get a result quickly. But if you want to modify the result on a small level, that’s basically impossible. You can’t change instrumentation, for example. And most musical prompts–in actual musical language, like »do this piano,« or »use these instruments,« or »make a ritardando«–tend to be ignored. The systems are good at recreating genres, and at generating lyrics, but not at musical detail.

 

I think there will be models that can do that in the future. I’m pretty sure.

 

At the moment, what interests me is finding out where the limits are and what happens when you push them. In a way, I’m interested in how I can break the machine.

 

There’s also the question of authorship: whose music is that, really? Who owns it?

 

There are no royalties for this music. And in Germany, from a legal perspective, there’s this concept called Schöpfungshöhe–which you could translate as a »level of creation.« Most courts would probably say that working with AI and generative music is a bit like working with samples.

 

It depends on the case. Do you just take a sample and place it there–and that’s it? Or do you take the sample and develop it further? Do you put work into it–and does something occur that becomes a work in itself, something you created?

 

I think it’s similar with generative music. If you generate material and then elaborate further and work with it, you can arrive at something that is still your music, from a legal perspective. But it depends, case by case.

 

Do you think it will lead to new musical genres or forms?

 

Yes–definitely. I’m pretty sure about that. What it’s going to be called, I have no idea.

 

Do you think we will have more avant-garde generative pieces as well?

 

I hope so. I tried a little with audio, and you often get stuck with something that’s either like John Coltrane free-jazz improvisation, or something that sounds like Aphex Twin or IDM. But it’s interesting to find out how you can morph these things.

 

Contemporary music already has this perception problem: many pieces sound »generated« to outsiders–even when they’re carefully composed with internal logic. If truly generative music enters the same space, the confusion might get worse. So maybe we’ll end up with an oversaturated market.

 

That’s already happening. Spotify is flooded with AI-generated music. I think there are even a couple of AI-generated tracks in the US charts. In popular music, it’s already a reality. I don’t know how sustainable it is. My best informed guess is that people will lose interest pretty fast.

 

In experimental music, there isn’t really a market–and that might be our advantage. We’re too small to be used that way. In a way, it’s a sanctuary. We’re still quite free in what we can do, and in what can happen within this field of new music.

 

There’s also a cultural split. There’s a huge audience for electronic experiments–modular synths, vintage gear, DIY scenes–and then there’s the Neue Musik/electroacoustic institutional world. They overlap, but they’re not the same ecosystem. Why do you choose to work primarily in the Neue Musik ecosystem–what does it allow that the other world doesn’t?

 

When I started composing, it was a conceptual decision to begin with computer music. I felt that if I wanted to make new music–contemporary music–I couldn’t limit myself to traditional Western European instruments from the last two hundred years, which were essentially finished in their development more than a hundred years ago. You just keep the string quartet as a string quartet, the orchestra as an orchestra–and I always felt that was limiting. So I wanted to use all kinds of instruments and technologies I could get my hands on. That’s why I decided to study computer music from the beginning.

 

And then, after a couple of years in the field–and also as a studio director–I made certain observations. One is that people tend to be more open to experimentation in electronic music. If you experiment with a string quartet, the group that will accept it, embrace it, or even stay curious can be relatively small. With electronic music, it’s different.

 

I think contemporary music is still part of the classical music market. Electronic music is more grassroots, more differentiated–and it’s also a younger technology. I observe more openness toward experiment there than in instrumental music, in classical acoustic music.

 

And it’s a spectrum, of course–from Berlin techno to Stockhausen, with deep connections and everything in between. I see tendencies in non-institutional, non-academic music that are similar to what happens in academic new music. Take ambient: people want to get rid of rhythm and danceability, and something like ambient arises. Or people want to get rid of melody and harmony, and they create noise music. It’s an analogue, in a way, to musique concrète and montage–but on an electronic level.

 

So the same aesthetic questions–the same discourse–exist in both worlds. They influence each other. And I also see people coming to the academy studio who don’t have the classical education of a composer. They come from other professions. They do electronic music as a hobby, but they’re serious, and they know a lot. At the same time, they often still have this high respect for the institution–the academy.

 

So there’s still a gap. And on the academic side, there can still be people who look down on things that aren’t institutionalized. But it’s not that simple.

 

There’s a big nostalgia factor. Electronic music »the old-fashioned way«–the idea of analog warmth, tape, »it sounds much better,« »vinyl sounds so much better«–there are many things there I wouldn’t agree with. But it’s interesting to see Hainbach popularity–that someone with a super niche topic can have huge reach and a big community.

 

That was striking to me: there are so many people interested in a niche subject. And if some of them are also interested in musical details and technical details, that’s great. It’s interesting to see how far this spreads–for example, Superbooth in Berlin every year, where thousands of people meet to play around with analog synthesizers, turning knobs and faders.

 

But you chose to be a new music composer for a reason, right? It’s another kind of attention and concentration, a different kind of listening. What’s your attitude to that?

 

For me, new music is not primarily about how it sounds, but about how it is perceived. In my experience, people who are deeply interested in contemporary music are deep listeners. They don’t use music as a personal soundtrack–to underline feelings or accompany what they’re doing. They’re interested in listening itself, in their own listening processes, and in detailed musical questions: how rhythm works, how harmony works, what a melody is, where the border lies between noise and sound, how musical form is constituted.

 

So I think people who are drawn to this music are careful listeners–maybe it boils down to that. And for me it’s the same. I was always interested in how music works, how musical processes work, how sound works, and how it interacts with me–with my brain, with my perception. I wanted to gain a deeper understanding of music in general.

 

If you could give the audience one listening instruction for the premiere–what would it be?

 

Maybe a very simple instruction would be: put away your phones. Just try to–because there’s so much distraction today. I see it at concerts too: people get distracted very easily. And if you don’t constantly pull on their attention, it becomes difficult. So maybe the instruction is simply: be patient while listening.