Created by Matteo Crivella, Markov Decisions is a generative audio system that combines random processes and sonic behaviours of matter. Data is created in Pure Data that drives the instrument through voice coil actuators putting the sheet of brass under stress and generating a tensional field in which the metal changes from one vibrating state to another, in relations to the input and its physical properties (mass, elasticity, morphology). The sound is the output that describe the behavior of the system, its vibrational state and the resulting actions.
“The generative process of data admits non-deterministic multiple future trajectories and defines the form and the dynamics of sound. The physical properties of the instrument defines the timbre. Low-level details about musical events are not specified, in favor of high-level decisions about setting parameters. After the set of rules have been specified, the system is left to work out the specifics of its data trajectory. The system integrates mathematical models, electronics, physical behaviours of matter and human behaviors as components of a larger cybernetic configuration. All this with the purpose to avoid choices regarding the composition, personal taste, bias and gestalt connections, in favor of a result that emerges from the choices made during the design of the system.” –– Matteo Crivella
Matteo emphases mathematics and calculation in order to objectify the process of music composition, thus moving away from a subjective composition processes. Whereas normally one would rely on own thoughts and emotions to influence the compositional decisions, in Markov Decisions the decisions are beyond human mind and control.
“My choices are in relation to my personal taste. Is it right other people have to undergo to the manifestation of my personal taste? Looked at from my ethical perspective it’s wrong. I don’t want to be the core of the process, but only a component aware that the true transcend the human and the only thing I can do is to dialogue with the outside. That make me feel safe from my narcissism. The roles of composition and instrument are definitively canceled in favor of a new identity instrument/composition totally programmed that doesn’t need to be played by a performer, human choices during the live performances are abolished, if you can program it, program it at all.” –– Matteo Crivella
Data is generated through the principle of frequency modulation, with a carrier low frequency oscillator and a modulator oscillator. Six first order Markov Chains make decisions for probability selection among data for variances in frequency and amplitude, generating changes in the vibrational state of brass that become tangible in terms of articulation, dynamic, loudness, timbre and brightness. A first order Markov Chain is a sequence of stochastic events where the next state of a variable of system isindependent of all past states, except the current state. The process consists of a finite number of states (in this case frequency and amplitude values chosen at random) and known probabilities of moving from one state to another. The Markov Chain Transitions Matrix (MCTM) are filled in at random, after that transposed in Pure Data. One for each operator; six in total.
Hardware includes voice coil actuators, amplifier and a Raspberry Pi 3, with 1 sheet of Brass 1000x2000x0.7 mm and 1 sheet of brass 1000x2000x0.5 mm, finally an Iron frame.