Automated transformation of large-scale musical structures in the form of perceptually smooth transitions is a non-trivial task. For the past few years, I have been studying and devising different approaches for creating transitions between sonic gestures (the smallest semantically complete sequence of sounds perceived by the listener as a coherent, target-oriented arc of energy, comparable to a musical phrase** in classical music), with the goal of achieving homogeneous yet evolving, and fluid sonic materials for my compositions.
Although the process I use so far works on an abstract level and is therefore applicable to any conceivable domain - including instrumental music as well as even non-musical fields - I have used it exclusively in my electronic compositions, which I categorically label as transitional music: music whose raw material is obtained mainly through transition-generating procedures.
I must note that by "transition" I do not mean the existing concept in the classical music sense (i.e. a passage composed to link one section of music to another), nor an internal transformation of a sound (of a recorded sound for example, using methods of spectral and temporal morphing, feasible with existing software tools).
Rather, I refer to a meta-transformation of the entire musical structure, consisting of large chronological sequences of possibly multidimensional (e.g. synthesized) sounds.
In order to conduct my research and my compositional work, I needed to redefine music in terms of time-series: a sequence of sounds is understood according to this definition as a time-discrete, equidistant sampling of events through time, thus yielding so-called data points containing significant features about each sound - such as temporal qualities, spectral content and degree of harmonicity/inharmonicity, localization and internal spatialization, intensity, or other articulatory properties, to name just a few. Other features either unique to synthesis or conceptually emergent from the basic ones are, of course, imaginable.
Such properties of sound are understood as different dimensions of such a data point.
This new perspective enables a more flexible application of the data-based transformation procedures I use. My research results so far culminate in an open-source software I have developed, which allows for the generation of transitions using arbitrary sources.
26.06.26
