logo

Name: Johannes Motschmann
Occupation: Composer, producer, pianist
Nationality: German
Recent release: Johannes Motschmann's AION 2, co-composed with AI and performed by the Ensemble Modern and the Johannes Motschmann Trio, is out via Springstoff.

If you enjoyed these thoughts by Johannes Motschmann and would like to find out more about his work, visit his official website or the homepage of his Johannes Motschmann Trio. He is also on Instagram, Facebook, and Soundcloud.

His work has been remixed by artists like Luke Slater and Chris Liebing.

[Read our Chris Liebing interview]



When did you first start getting interested in AI in relation to music? What is the appeal of this field and what, to you, are interesting and/or successful forays?

I was asked by Christian Fausch and some members of the Ensemble Modern to create a composition that was composed with Artificial Intelligence. For us composers, this distinction from methods of conventional algorithmic composition is very important, because automating the composition of music is nothing new, composers have been doing it for decades.

So the question for me is not whether software can create coherent pieces of music - it can do that anyway - but rather whether it can spontaneously, reactively, in real time, invent material that fits the music that is being played. And even more importantly, perhaps, whether it can convey emotions.

So it's about more than algorithmic composition. It's about an interaction of human and machine on stage. Acting as planned as a composer and reacting as fast as a trained jazz musician improvising to create a touching musical statement in real time. This intermediate field between composition and improvisation between planning and spontaneous execution is the actual focus.

AI has been used in many different forms of art, from poetry via contemporary art to videos and music composition. Where do you personally see the biggest potential?

I don't think it's about a specific genre, but about a differentiated and appropriate approach to technology in each genre. The most important thing is to get away from childish comparisons of who can do it better or faster, humans or machines. That will get us nowhere. Instead, we should ask ourselves how we can get the machine to support something new in interaction with our ideas.

There are things we would never try out of convenience. Through the sheer speed of the computer, we gain insight into what is possible with our material without spending weeks writing down variations. You can try more, discard faster, and start over, that's the appeal for me. Translating quantity into quality and finding strategies for that with the help of software.

The term AI is an interesting one when it comes to creativity. What, would you say, does intelligence mean in relation to music?

The problem is that our definitions of intelligence are generally very vague. What is intelligence and what is not is ultimately a question for philosophers and neurologists.

As a musician, I would always answer pragmatically and measure methods of composing, of thinking in music, by the results. If, with the help of artificial intelligence, I succeed in making music that appeals to people and touches them emotionally in the same way as music that I wrote in ink on a sheet, then I've advanced.

What was the spark for working with AI for AION?

I don't want technological innovations in the industry to be used against the artist.

For example, when streaming services create music through AI in order to not have to pay royalties to composers, which has been happening for a long time. Our task as musicians is to take composing back from those people who produce music not out of artistic interest, but out of purely economic calculation.

Here, too, a paradox: I would like to see through the strategies of composing without composers in order to then do something with this technology that only a creative person can do. Indeed, I firmly believe that our world is a better place when composers compose, photographers photograph, and designers design. Especially in a world that makes us believe that everything can be done with little apps on our smartphones, there is a danger that art will be devalued according to the motto: anyone can do that.

Of course, not everyone who has a good camera in his smartphone is a great photographer. We just have a lot more photos than in former times. It's the same with music, which can be produced by software at the touch of a button. We have to know the strategies ourselves, see through them and ultimately do something that no software can: be incalculable in the literal sense of the word!

You've obviously worked with algorithms in music before. Was there nonetheless the feeling of treading new ground here? How does AION expand beyond past computer-aided composition approaches?

In AION, I am concerned with the concept of eternity and what it means when there is no longer a time limit on music, when it becomes inflationary, to put it negatively. For example, if I have a 10th symphony composed by Beethoven that suggests that there could be 1000 more like it, I quickly get bored.

The important thing is always to keep in mind that the lifetime of composer and listener is limited, and the realization of music as time shaped by sound must take this into account. A machine that per se makes no distinction between producing one minute of music or 1000 hours of music is dependent on humans.

Paradoxically, the strength here is the weakness: limitation of time and an economy of applied means, are important categories for composing. Mass and speed will not achieve anything with the audience. After all, we do not judge a composer by whether he has composed a lot or quickly, but by whether his ideas speak to us.

The prerequisite for music created by artificial intelligence to be taken seriously depends on the quality of the music. And that in turn depends on very subjective judgments. The difference between conventional methods of automating music and new approaches lies exactly there: the free choice of humans over musical material.

Not simply to produce music, but to create music that could also be the result of human deliberation - even if it brings something significantly new and does not just imitate tried and tested styles.

How, exactly, does the AI that you used for this project, create harmonies, rhythms and melodies? Can you tell me a bit about the process of extracting materials from it and integrating them into your work?

I worked together with Thomas Hummel (Experimentalsstudio des SWR) on the software and he wrote the code. So I can only describe on the surface what we did: I can put it in one sentence. We tried to find out how I compose myself.

We humans often proceed according to a plan. According to the algorithms in our head, according to experience of music and music-theoretical knowledge, and through something that is intuitive, a play instinct a trying out. We have searched for each of the methods of thinking out music an equivalent that we can simulate with this software.

There is one very important distinction: what does the AI "listen to", how much music is in the buffer. If I played Chopin all day, the results will sound like Chopin. Just as we are influenced in our composing by the music we know. That would be the very classical area of data processing. The more music there is in the buffer, the more differentiated the answers you get.

But the other side is the procedure with the material, and therein lies perhaps the music-theoretical knowledge of Thomas and me. Namely, to create music immediately even with little input, without Chopin and Mozart being in that buffer. So it is a hybrid of data processing as in almost the AI and music theory translated into algorithms. One side limits the results in terms of style, the other one delimits them again for more abstract and less connotated results.

There are methods for how strongly autonomous the answers of this software turn out, completely independently of which music is in the buffer and how much of it.

Working with AI allows for glimpses at non (traditional) human musicality. From your own experience, in which way is it different from our human musicality? In which way did this project lead you to shores you would never have visited otherwise?

I would not be patient enough for some experiments. And I can select very strictly because I have an infinite amount of good-sounding musical material.

Sometimes you want to use something that you have painstakingly worked out because it takes energy to write it down. I want to focus my energy on the content and not on redundant steps. But to be honest: so far, this only works quite poorly sometimes ...

Purely from an aesthetic point of view, how do you rate AION compared to your earlier work?

It is more free in dealing with role models. From Karl Kraus comes the beautiful sentence: “Der lesende Dichter macht sich verdächtig”. (“The poet who reads is suspicious.”) With AION, a lot of creative power went into processing music by other composers.

I tuned the AI together with Thomas to different styles. With music by Bach, Mozart, Rachmaninoff or Boulez. If I find my own way of composing in this software, I could better distinguish myself between things that arise from nothing and musical passages that have a classical model. Ultimately, a memory of other music evokes a different style.

By the way, this is something that I have noticed in almost all composers: even in Mozart there are passages that could almost be from Johann Christian Bach, and Cherubini sometimes sounds a bit like Beethoven (or vice versa?) This can then perhaps be described as contemporary style or fashion, and there are hardly any composers who did not respond to the stylistic characteristics of their time.

And in the same way, however, there are these tipping moments in the Hammerklavier Sonata by Beethoven or the B minor Sonata by Liszt, which have absolutely nothing to do with these fashions. They seem to come from another star. I find these two sides exciting. The familiar (also in other music) and what one seems to have never heard before. I think good music is somehow between these poles, of comprehensibility and incomprehensibility.

Even if AI will not entirely replace human composition, it looks set to have a significant impact on it. What does the term composing mean in the era of AI, do you feel? How does it change perception of ownership of a work?

The composer Johannes Kreidler has shaped the term "composing with music" and I think he means that composers may and should take into account the ubiquitous availability of musical material in our time. This touches on issues of copyright and the question of the Romantic role of the composer as a genius who creates incredible things out of nothing.

However, our romantic notion of composing undercuts the great amount of craft and experience through the music that sounds around us every day. I would like to combine the discussion about artificial intelligence in music with the question of whether we have already outlived the classical role of the composer who, as a creative human being, creates music virtually out of nothing. Then it would be easier to deal with traditions or to break with them ...

Can you see a future where AI could make aesthetic judgements as well?

It can help expose a lot of soulless music by people who don't put much effort into composing. As I said before, composers themselves often follow a pattern, have their own rigid algorithms in their heads, and somehow extract musical material without any creative effort. Such music would then no longer keep up with the results of artificial intelligence.

Let's see the technology as a possibility to better understand our own actions and at the same time to become unpredictable in our music.