Bit-makers - with Giulia Lorusso

BINARY PORTRAIT

INTERVIEW

COMPLETE INTERVIEW

“Composer-researcher”. Do you recognize yourself in this definition?

I am a composer, but I wouldn’t dare define myself as a researcher. Except maybe in the sense that when you're an artist— or at least a composer, because I don't like to define myself as an artist either— you're always looking for something.

But I am not a scientific researcher, in the sense that I have no scientific training. Not yet, you never know! And I'm not in the research business either.

Nevertheless, I am currently involved in an artistic research project, in a duo with a very good friend who is a computer scientist: Alessandro Rudi. He comes from Rome, like me, and works as a researcher at INRIA. He studied in Genoa and at MIT in Boston, he did a post-doctorate in Paris and he is now a researcher and professor at the Ecole Normale.

I’m lucky to have Alessandro as a friend. We met up in Paris, and we wanted to work together. First of all because we got along very well. And also because there is an kinship between the way of a researcher thinks and the way a composer thinks: we are always in research and reflection!

We decided to propose this project together. I have always been fascinated by the scientific world, by research, and in this case, by the field of computer science, artificial intelligence, and related technologies.

So do you think you two are complementary? Are you looking for the same thing in different ways?

Yes, we used to meet on Sunday afternoons, I had pieces to work on and he had mathematical problems to solve, or articles to write. We both found ourselves in the same state of perpetual reflection, being caught up in something that affects us beyond simple work, that enters into personal life. It must be the same for a writer, or a researcher in any field.

Parallel to this affinity, we shared a lot of ideas and questions, artistic on the one hand and scientific on the other.

He always talked to me about his research, the algorithms he was developing, and it so happened that at one point we presented a project here at IRCAM.

What we lacked, and what Ircam is offering us now, is an interface between the two of us. He is a researcher but not a musician. And I am a musician but I have no scientific training. We could communicate, but in reality we needed a structure, we needed to work in a team with other researchers who could help us.

I'm talking above all about human structure, that is to say, about setting up a team work, and about being able to get in touch with people who already have this double experience of music and artificial intelligence.

I think this is one of IRCAM's strong points, because a lot of research is being done in this field, and artificial intelligence is in fashion! What Ircam offers, which is if not unique, at least a rare case, is to mix music with these computer technologies. There are a lot of people here who have dual training, which neither of us has. 

So I'm not a researcher, but I'm interested in this field and I'm trying to learn these computer tools myself, to get into them as much as possible, to go as far as I can. Even if I don't have this double training like Andrea Agostini or like Daniele Ghisi who has a doctorate in mathematics.

So you consider yourself a musician, a composer, curious about your environment.

Yes, I do.

You told me about Alessandro Rudi's career, what is yours?

Until now I have had a rather traditional training. I say "up to now" because I think we are always reinventing ourselves, training in different fields.

Traditional in the sense that I first started with the piano. And then I continued with composition. First in Rome, in my home town, then in Milan to train with Alessandro Solbiati, whom I consider to be my first composition teacher. There I did a degree in composition where I also studied piano, which I stopped during my Erasmus year at the CNSM in Paris.

I did a Master's degree while studying at IRCAM. That was my discovery of electronic music. In Italy, unfortunately, the teaching of composition and that of electronic music are still very separate. They are two different diplomas.

This is amazing, because there are a lot of Italian composers in electronics and electroacoustics!

Yes, they were trained in France! (laughs) I’m joking! There are a lot of composers that I find very interesting who were trained in Italy. For example, I am thinking of Agostino Di Scipio who taught in France, I think. Or Michelangelo Lupone, who also has a very original and personal approach to electronic composition, different from the generation that trained here at IRCAM.

Finally, to keep honest about the situation in Italy: we have very good teachers and a good teaching in electronic music but we don't have the means, nor the infrastructure. We had classes but no software, no computer labs or studios. And it's difficult because, from my point of view, electronic music is not taught in a theoretical way. Of course there are scientific and acoustic bases, but it's very important to have material support, studios, the possibility to do experiments on the computer.

So in Italy, this deficiency is not the fault of people, but of infrastructures. We have more and more, but not like in France, let's say.

In your case, when you left Italy you were more into traditional, instrumental composition. Was the discovery of electronics more a coincidence or a desire or a need?

No, it wasn't a coincidence, I had this desire, I had this curiosity. As with many composers my age, there's a very deep connection between acoustic and electronic approaches to composition.

For example, when I compose mixed pieces, I often compose the electronic part first and then the acoustic is mixed in with a game of back and forth; but the approach is the same.

Would you say that your relationship to electronic writing is the same as your relationship to instrumental and acoustic writing?

Yes, increasingly so. I see it as a refinement in my approach to sound and sound material.

The year I spent studying here has given me a lot of food for thought. It has given me a lot of input in terms of instrumental writing and thinking about the creative and compositional process.

I like to say that working with electronics is a kind of mirror. It's subjective, but I find a more instinctive side to it. When you work with electronics, at least in my way of working, you immediately have a tangible result, and that's what allows this instinctive relationship to the sound material. It allows you to manipulate it as if it were physical matter, as if it were sculpture. Conversely, instrumental writing requires much more control and is much more spread out over time.

Again this is very subjective, because other composers have a much more algorithmic approach and can spend months designing the processes, which is another approach.

When you talk about mirrors, do you mean that having an immediate result and shaping it to your liking in the moment, that says something about you?

Yes. When you stand in front of a mirror, you immediately have an image. You can stop, look at it in detail, let things surface...

It is this immediacy that allows me to generate sound material, to deform and rework it, very quickly and by successive rewrites.

I also do this with instrumental writing, that is to say by a kind of parody. I write, I rewrite several layers. It obviously takes months with instrumental writing, whereas it's very fast on the computer!

I often start from my own pieces and I transform them, I tear them up... I record elements and I improvise with the computer, then I re-record. It's a work of transformation that can lead very far! That's the power of computer work.

In your career, are there any important events or figures that might emerge?

I meet a lot of people, and it's the meetings that shape us, more so than the educational institutions. I've met several composers who have left their mark on me. I can name some of them and that doesn't take anything away from those I won't name here: Pierluigi Billone, whom I've only met five or six times but who is very generous and instinctive. I have a lot of affinity with his way of thinking and of conceiving composition, he influenced me a lot. Or rather he opened up perspectives for me, which is the most important thing. I wouldn't say that my music resembles his, but he opened up horizons that have allowed me to be where I am today. Another important person is Marco Momi, who also opened up other horizons for me through his advice and his music.

There are also a lot of indirect encounters I have made, reading scores and books. At the moment, I'm more and more interested in other fields and especially in the plastic arts. I read a lot of Nicolas Bourriaud who doesn't talk about music at all! He deals with visual arts but it's a process that's still going on and that opens up other perspectives for me. I've also just read La révolution digitale dans la musique by Harry Lehmann. I consider all of these as indirect encounters.

There are many others, of course. I've only mentioned men, it's a pity!

So I will also mention Clara Iannotta who is an Italian composer and a friend, above all.

Can you tell us about your work with Alessandro Rudi?

Our work has just begun. It revolves around the application of a specific Machine Learning algorithm in the field of music. It is an algorithm that Alessandro developed with Francis Bach and his team at IRIA and Carlo Ciliberto who works in London.

This algorithm is called "Localized structure prediction" and has found very nice applications in the field of images.

Its principle is to recognize emerging patterns in a given image, then to search for and restore both local and global coherence.

The metaphor we propose is that of Google Deep Dream which is a visual experience. The idea was to train the machine to recognize various images of plants and animals. We build the dataset, we create couples. Then, by providing it with an image that doesn't fit into its dataset, the algorithm will recognize emerging patterns close to what it already knows. For example, it will recognize shapes of animals or plants in images of landscapes that do not contain them. Then, beyond the recognition of these emerging patterns, it will reconstruct a global coherence. For example, he will be able to recognize the line of a dog's ear in the image of a leaf, and then reconstruct an image of a dog. It is a little simplified and approximate but it is a way to describe the functioning of this algorithm.

This type of calculation is mostly used in images, for example to obtain super-resolutions, by generating several pixels from a single one. It is therefore possible to generate high definition images from low quality images. This is also useful for colorizing black and white images. This is called a generative algorithm, i.e. it adds missing information.

This algorithm has not yet been applied to sound. For this reason alone, it may be interesting to see what results are obtained. It also has a peculiarity in that it does not use neural networks. So the interest is also scientific.

From a musical point of view, there is obviously an interest in working with this type of tool. The first idea is to generate coherent musical textures from any input. And even before talking about music, we are working on the audio signal: from a sound input, generate a texture.

Artificial intelligence has been the subject of many experiments in music generation: we make the computer write music in the style of Bach or Mozart, or we use it to generate scores, that is to say in symbolic generation. That's not what interests me; I prefer to stay in the generation of timbres and textures.

In my artistic research, what interests me is putting these algorithms in relation to external inputs, in this case environmental sounds. That is to say, to work in the field of concrete sounds and put them in relation with the musical field, so that the algorithm can generate music from these concrete materials.

What I would like is for the system to be able to do hybridization. It's not just transformation, taking a subway noise and making music with it. Rather, it would be a matter of bridging that sound and the music produced.

In fact this bridge already exists. Raymond Murray Schafer evokes it in his book Le Paysage Sonore (which dates back to the 70s!). According to him, there is always a relationship between the world of sound in which we evolve and the music we produce, at a given time, in a certain era.

That's very true for me. I'm aware that I'm very influenced by everything around me and especially by the sounds around me, and that I have a very particular listening directed towards them. When I take the subway, there are sounds that I find very interesting, textures, and even musical gestures, phrases that I recognize there. My daily sound experience influences my work enormously. At all levels: at the level of temporality, at the formal level.

Are you afraid of being taken for a composer of concrete music?

There is a relationship that is even deeper, which is not just the sound of the subway that can sound like a multiphonic clarinet, for example. It would have more to do with the speed with which we perceive the sound or the soundscape around us; the influence is on temporality, form, and above all on perceptions.

As a composer, the perception of time is something that interests me and it is especially in this field that the external sound world influences me. We talk a lot about timbre, texture... it's also valid, but for me these elements are always in relation with "the whole", and especially with the temporal dimension.

The starting point of our research was to put in relation external sound and music. But I wondered what would be the sound image we could have of our world today. And I don't think we can reduce it to the soundscape that surrounds us, because it is superimposed on the enormous amount of music we are immersed in. It's as if we're immersed in a gigantic dataset of music that's constantly surrounding us.

My first artistic project is to try to work with this totality of sound data. By making a selection, of course, but taking into account the flow itself, which is constantly changing.

We want to use this research in a piece for an electronic ensemble. We have a commission independent of the residency, but we will try to implement the results we get. My idea is to create a relationship between what happens on stage, and therefore the written piece, and the flow of data that is constantly changing. This flow will come in particular from the web: there are datasets of videos, images, sounds. These datasets are open-source, so they are constantly changing. They give a sound image of our world.

As I was saying, this type of machine learning has been very much exploited in the visual field. I have learned a lot and I am fascinated by this kind of technological research, but in which the visual field is privileged.

And little in the way of sound?

Not little, let's say less.

What is certain is that it is easier to see a "visual" image of the world. So what would be the sound image we could have of it? That's what interests me.

How could I represent from a sound point of view our world today? This is my main question.

Do you see a social dimension in wanting to give an image of the world? Or is it purely artistic?

I understood that the impact I could have, or let's say what I could contribute, was mainly artistic. And I feel this need, that art should have an impact in society, a stronger social impact. I miss that. Especially in the kind of experimental art, as I do and as others do...

I wonder a lot about the impact I can have in the world I live in, and I feel a little frustrated. Does my work contribute anything? Outside of the institutional setting of course, because I feel that what I do may only stay or have an impact in an institutional context. That's why I'm wondering if my idea for a play is not a bit ambitious?

Is there a way to use an artistic project, a piece, as an opportunity to broaden this context? To try to create a bridge between a reality that goes at an extraordinary speed and what is happening in the concert hall? To create an environment that allows a connection between what happens inside and what happens outside?

Do you have a message to pass along in this connection?

Yes, in the piece I'm working on, I have textual material that I want to use as sound material, which includes for example extracts from the Livre du comité invisible. These are excerpts that I have selected and which have a fairly strong social critical content. Social but not political, because I believe that art and politics must remain separate, so I don't want to take a position.

The text is already a way to go in that direction, but I am looking for other ways, even if I don't have a solution right away. It's something I want to explore in any case.

I know that the impact I can have with a piece played in a concert hall is quite minimal and touches a specific audience, usually quite cultured. But it's a first step, maybe?

One of my concerns, and this is what makes working with artificial intelligence so interesting, is to put art in relation to current events and the reality that encompasses this flow of data that our world represents. An artistic work, whatever it is, must not be detached from reality.

Can you tell us about your work? What are your sources of inspiration?

As a composer, I am above all concerned with sound. I have talked a lot about sound matter and this reflects my interest in it. I also talked about concepts because my project is still at this stage, but I'm not really very conceptual.

My starting point is often the sound itself. The physical, human dimension of sound interests me and influences the way I deal with sound matter.

And again, the question of temporality. It's a work on perception; I start from a sound and I try to conceptualize it to create a perception of temporality.

The musical gesture seems to be an important component for you.

That's quite true, it is an important component. My latest piece for example is a piece for two performers and electronics, made in collaboration with a visual artist. It's all about breathing, I use the percussionist's breathing and the performer's breathing.

In the piece for accordion that I wrote for the Cursus too, this dimension is important. It's a very physical piece for the performer!

Having said that, it's difficult for me to isolate elements and say "the gesture is important" or such and such a thing is important, because I write and rewrite a lot and I see it as a whole.

In your music, we can notice an opposition between the continuous and the discontinuous, as if textures were built from dissociated elements and then broken. Is this an interpretation you approve of?

I don't particularly see it as an opposition. I see it very much as a movement of energy. Or, metaphorically, as a process of depletion. I conceptualize a lot by tactile metaphors. I think of sound in terms of roughness of consistency, density...

It is therefore also a question of perception: you distinguish between the objectivity of the material and the perception that one can have of it.

Yes, it is this physical characteristic, this "plasticity", but in perception, and therefore in relation to the temporal axis.

I am more and more aware of this question of temporality, I have talked a lot about it. Unconsciously or consciously, we work on it whatever happens, and I feel more and more the importance of this concern.

Classically, the teachings emphasize harmonic, timbral, formal questions... but these are points of view! What changes the result is the perspective that one decides to take as a composer, to tackle an idea or a sound; one will not have the same result by approaching it from a harmonic or timbral point of view. Like the same object that gives different shadows depending on the way one illuminates it.

And yet we have to take into account the whole thing. Even when looking at it from a temporal point of view you must not lose the harmonic dimension. The temporal dimension is important to me because I feel more and more like a key to understaning contemporaneity.

Interviewed by François Vey.