“It really comes down to intention and how much you want to reveal and how much you don’t want to reveal, and it’s like a delicate walk between the two. Often times I rely on the material and the memory of the material doing a little bit of the work for me. I also rely a little bit on the subject matter to do a little bit of the political work for me. And then I also give myself the freedom of abstraction to go wherever I need to go so that I am not a spokesman for something.”
Mark Bradford from an interview on the exhibition Mark Bradford: End Papers at the Fort Worth Modern.
https://www.youtube.com/watch?v=Y3UbK5H5UuI
This statement is so complete in solidifying the artistic purpose of abstraction in relation to the material and the political dimensions of the artwork that I could have stopped here without further elaboration. Indeed, I would not have been able to describe with words the essence of abstraction within my own practice nearly as accurately or eloquently.
Delimitation
I will only add a few words as an introduction for the reader to my music and this artistic research project by means of negation, in a way that will hopefully set the stage for what is to follow. The fields of improvisation, electronic music and algorithmic music within which my project is situated are so broad and the practices within them so diverse, that, in order to define in a concrete way what is, one might benefit by defining what is not.
I have often that when I talk about my project without clearly outlining from the start the elements of abstraction inherent in my methods and artistic processes, false assumptions arise. Without this contextual framing, my discussants tend to assume that I am pursuing something different than what I am actually doing, and consequently expect a radically different artistic outcome that the one I produce. And, once these false assumptions are established, further misunderstandings naturally follow. Therefore, I need to keep in mind that aspects of my practice and this artistic research project that I consider to be trivial or self-evident are not necessarily obvious to others. For these reason, I will attempt to clarify these elements early in this text.
As the reader proceeds, it is important to keep in mind that I am describing only my own aesthetic perspective, which is the driving force behind my music. I only present how I personally experience and relate to improvisation, sound, interaction and other related concepts. These reflections should therefore be understood as applying solely to this project, and not as general statements intended for generalization or extension beyond this work.
Aesthetics of Interactivity
I am an improviser and electronic music composer. It is not uncommon for musicians who play both acoustic and electronic instruments to be interested in systems that facilitate real-time interaction between the two worlds. Often times, in these real-time interactive systems, the acoustic sound is manipulated by electronic means or triggers electronic sounds. The most common assumption that people make when I say that I am a pianist who also composes electronic music is that I work with these types of real-time interactive systems.
Nevertheless, this is actually something that I have consciously avoided in my own compositions. And there are two main aesthetic dimensions underlying this choice. The first aesthetic dimension is related to my desire to avoid linear causality. This is a central element of my project and will be introduced later in this text when I will describe the notion of ‘one-to-one’ mapping, before being elaborated in greater detail in later texts. The second dimension stems from an aesthetic choice that is central to all my music: namely that acoustic and electronic sound are not discrete, and therefore no hierarchical structure can emerge in which one controls the other.
I am originally a pianist who later was introduced to electronic music and modular synthesizers. My electronic music composition informed my piano playing and I started placing objects inside the piano (what is called prepared piano) and using extended techniques in order to imitate electronic sounds with the piano. At the same time, in our duo with synthesist Egil Kalman, we explored deeply this relationship. In that setting, I attempted to play piano in a way that resembled electronic sound, while Egil played the synthesizer in ways that imitated acoustic sound, blurring the boundaries between acoustic and electronic sound to the point that the listener was unable to distinguish between the two. For me, acoustic and electronic sound therefore interact perceptually by coexisting in space and by means of abstraction, as two equal voices running independently and in parallel rather than in a hierarchical manner.
These aesthetics of real-time interaction between acoustic and electronic instruments in my work have been extended to the interactions between sound and other media, and more generally to human-machine interactions. In our composition Athroa for the same duo (Efstathiou/Kalman) and a light installation, applying the same idea as above, the light installation was treated as another layer within the improvising ensemble and as an equal voice running independently and in parallel - a third member of the improvising ensemble.
Bringing this background to the project, my work with machine learning would naturally continue in this path. Creating real-time co-improvising agents that receive my real-time piano playing as input is not within the scope of this project - at least in the present moment when I am writing this text. This is an aesthetic choice and not a dogma, and I allow things to change and evolve as my project develops. I instead aspire to create connections between machine learning algorithms and my artistic practice on a conceptual and abstract level. Machine learning algorithms will be incorporated in an applied manner within my compositional practice for sound generation and other processes, to the extent they are relevant to my artistic practice, broader artistic questions and aesthetics.
This project is not only limited by my aesthetics but also by its time scope. I am part of two working packages within the MishMash Centre for AI and Creativity. The scope of these two packages combined is very large, and the field of machine learning is both broad and emerging. So, in order for my project to have any meaningful contribution, it must avoid the pursue of all possible fields of interest and instead be limited and precise.
One-to-one mapping
As it will become clearer in later texts, I follow an improvisational tradition where the improvisers interact by following parallel and overlapping paths rather than linear call-and-response patterns. This way of playing allows for abstraction and invites the listener to “make sense” of the music by creating connections between the different voices. My aesthetics of interaction are therefore informed by an improvisational experience in which interaction is both indirect and multilayered.
For this reason, it does not feel natural for me to work directly with one-to-one translations between acoustic-electronic sources or different media. By one-to-one translation, I refer to situations in which one medium directly and deterministically mirrors another. One might imagine, for example, a dancer moving with a music that is tightly locked to her every single gesture. In such a scene, two signals that are perceptually reduced to a single one and the spectator experiences the interaction through the lens of deterministic causality.
This notion of one-to-one translation or mapping also extends to the practice of sonification, the mapping of a mathematical object -such as a function- to an audio signal. I find that the causal transparency of the one-to-one mapping removes many degrees of artistic freedom, as it favours a mono-signal approach over a more complex contrapunctal relationship between media and signals. In doing so, one-to-one mapping limits the potential of abstraction and reduces the audience’s ability to create associations through speculative, imaginary, and cognitive means.
Going back to Bradford’s words, it is indeed a question of how much you want to reveal, and when revealing too much, one does not leave quite enough room for the artistic freedom that comes with abstraction.
These ideas are, of course, not original. There is a long tradition in experimental music, dance, cinema and video art that explores similar approaches. For example, in Xenakis’ Polytopes -monumental multimedia works for light and sound- the visual and sonic elements do not control one another, they instead create a counterpoint to each other.
“In contrast to the complex light program the tape-music is very simple, providing modulating timbres, varying pulses which are heard in counterpoint to the rate and density of lights constantly going on and off.”
https://www.youtube.com/watch?v=XK1KjtetCsg
Extended techniques
I have already mentioned above my practice in the piano that involves using objects inside the piano board to create sounds that resemble electronic music sounds. These techniques have been systematically developed over the past eight years and were well established before the start of this artistic research project. These techniques are now some of my basic tools that feel as a second nature to me and, although they are always a work in progress, developing the techniques is not the main goal of this project. The purpose is to demonstrate how the embodiment of these techniques is possible through the concepts presented within these pages.
What is important for the project is not the particulars of individual techniques, but their precarity - their affordances and limitations, their stochasticity and the complexities they give rise to.
I find that the interest in the techniques themselves often distracts from experiencing the sound. My objects are colourful and visually engaging and involve intricate gestural performativity, which I often sometimes feel comes between the listener and the sound. Ideally, I would like my music to be perceived without the distraction of my physical presence on stage. It is best experienced as electroacoustic music, where the sound source does is secondary and the sound itself remains at the forefront. The absence of videos in the following pages is therefore purposeful, placing the sound at the centre of the listener’s attention.
I am, however, well aware that the techniques I use may be of interest to other pianists and to readers of the final exposition. For this reason, the techniques will be presented in a separate section of the final exposition. This section will include descriptions of the techniques and their respective tools, accompanied by video demonstrations.
Method
This project seeks to understand the complexities of my artistic practice by placing it in dialogue with ideas drawn from other complex systems. Although initially situated in relation to dynamical systems in general, the project has since become more focused on machine learning systems. To make this connection possible, I pursue a deep understanding of machine learning algorithms. My work involves a substantial educational component that includes studying computer science texts, following online course materials and programming with standard machine learning frameworks. A separate section of my final exposition will be dedicated in sharing these resources.
I initially assumed that studying machine learning texts and reflecting on the relationship of machine learning algorithms to my artistic practice would constitute two distinct and separate processes. Over time, however, it became clear that these paths could not be disentangled. My engagement with machine learning is inseparable from my artistic practice, and reading technical texts inevitably triggers reflections on my own work. This reciprocal process has developed into a central method within the project. As I read about machine learning algorithms, ideas relates to my artistic practice frequently emerge. Recognising this recurring pattern, I began to systematically document these reflections as they arise. Several of the ideas presented in the following texts originate in these initial reflections.
