Preliminary Study
For Percussion Tro

Diarmid Flatley

 
 
 

Preliminary Study for Percussion Trio is an absurd title for an audio-visual work that does not contain percussion.

Preliminary Study for Percussion Trio is both an artifact of process and an end unto itself. The piece is the result of compositional sketches for percussion trio. 

The initial purpose of the Preliminary Study for Percussion Trio was that of a maquette. An audition of stochastic processes which generate formal structure. The stochastic model under investigation in this work is the Markov chain.

The Markov chain defines the likelihood of one sonic event following another. It determines each performer's role within a composite meta-gesture. It guides the evolution of musical material. 

The Markov table indicates one of three roles for each performer per sonic event. These roles are either a staccato attack, a sustained sound, or a silence (represented in the table by"x", "o", and "-").   

Preliminary Study for Percussion Trio demonstrates how easy it is for a process to derail itself. The goal of the work was to test the aesthetic quality of probability structures. Yet, as I began testing these structures, the relationship between the work and the percussion trio became tenuous. Beyond tenuous: unimportant.

Square wave oscillators produce the sounds in the piece. Each state in the Markov chain defines parameters for the oscillators. These parameters include frequency, pulse-width and duration. The progression through a series of states results in an evolution of sonic texture. That is to say, on some level, the formal structure of the work is the stochastic algorithm itself.

Recordings of this stochastic process unfolding are the source materials for the work. These materials are subject to intuitive interventions. Material, fragmented and piled upon itself, gesturing towards overwhelming density. The texture thins near the end as the source material becomes exhausted.

The visual aspect of the work maps audio analysis data to pixel information in real time. Audio analysis data, polled at frame rate, determines the red, green, and blue values for each pixel. Simple audio-reactive video compositing algorithms, further mutilate the image. The resulting video is intrinsically linked to the audio. However, it is more a visual commentary on the audio than a visualization of the audio.