organic_algorithmic_composition

Selected Projects

2023

Pebbles to Clouds_on Web

AI + Web Audio

Organic Algorithmic Composition

AI + Composition

2023

Organic Algorithm with AI

→ AIMC2023 Proceedings on Pubpub

AIMC2023, Sussex University, Brighton, UK

Performance Proceedings AI + Music

» Organic Algorithmic Composition « is not just a musical composition. Instead, it is a sonic exploration that raises the question of whether sounds can self-organize within a given system.

A continuous chain of musical possibilities is generated with a simulation of one of the slime molds, "Physarum Polycephalum.” The stages of generation and modulation will be implemented with Physarum’s decision-making method with a multi-agent model. In generation/modulation stages are operated with several agent-based algorithms to get a correlation. So that It generates non-linear musical behavior in each phase, creating unique musical patterns.

Simulations of Physarum (a genus of slime molds) and Rave(Callion, Phillipe 2021) Autoencoder are played in parallel. They converge, diverge, or neglect and oscillate within a system. This local oscillation creates unique sound patterns: harmonized feedback, smearing, and displacements. Physarum simulation data transforms into sounds. It controls sound oscillation, modulation, and the nonlinear musical behavior of each phase, creating unique musical patterns underlying the principle of Self-organizing in music that Interwining Organisms and Modern computational functions in AI.

For the AI, Ircam's Rave Autoencoder will be implemented for sound processing with live performance to achieve Sound's self-organizing like organisms. The interconnection between the simulation of physarum and Rave under the principals of nonlinear, bottom-up approaches in a music compositional system explores the new auditory domain. It unfolds the possible connections between humans and non-humans, objects and organisms, and reality and non-reality. It intertwines machine algorithms with sound and uses the principles of an organism's behavior to expand the possibilities of self-organizing sound and the organicity of sounds in space.

Image source : https://www.quantamagazine.org/slime-molds-remember-but-do-they-learn-20180709/

Physarum is a large Amoeboid Myxomeycete organism. It adapts its body plan during its complex life cycle to various environmental stimuli (nutrient attractants, repellents, hazards). Physarum has drawn considerable attention from researchers due to its simple cellular structure and decentralized control system.

An early model of Physarum was focused on individual biological aspects of its behavior, most notably the generation, coupling, and phase interactions between oscillators within the plasmodium. More recently, the overall behavior of the organism has been modeled in attempts to discover more about its distributed computation abilities (Jones 2015). Over the past few years, there has been a surge in research exploring the computational capabilities of Physarum, primarily driven by Toshiyuki Nakagaki (Nakagaki. T) (Jones 2011). Nakagaki has researched the subsequent analysis engaged Physarum’s behavior expands to the computational abilities. Which are solving path planning problems (Nakagaki, Yamada 2000) and combinational optimization problems. (Aono, Hara 2007)

Physarum simulation implemented in the multi-agents simulation with GLSL(OpenGL Shading Language) to expand the ways of synthesis and underlying principles of self-organizing in music with non-linear, emergent behavior. According to the multi-agent model in this paper (Jones 2015). Simulation of Physarum, not just using it to generate visual patterns, but local oscillation patterns are directly translated with sounds. Each Agent has a “window” (square in simulation). It can monitor the direction of specific agent behavior and specify the adjacent's local oscillation data. The real-time data of the agents generate the sound with the 2048 sampled wavetable synthesis method. So that Within a simulation system, each specific agent’s local movements and patterns, not just the result of the visual simulation, each local patterns directly connected with the sound generation stage.

Multi-agent simulation of Physarum Polycephalum by Jeff Jones, Jones 2011

Generated Sound from simulation and Rave AutoEncoder play in parallel within a system. This paper (Canonne, Garnier, 2011) illustrates the nonlinear behavior in free improvisation and demonstrates the mathematical loads derived from thermal physics for nonlinear behavior in improvisation on a stage. In this manner, a derived real-time correlation of each agent can be represented as specific values. The values control for sounds between the sound of the Physarum simulation and the Rave (Caillon, Philippe 2021) Autoencoder. And the amount of negative feedback in a system. This correlation and the combination of the simulation data, such as the agent speed, direction, and density, will be used as the deterministic factor for the music forms.

Rave Autoencoder will be trained with Human voices, artificial hardware sounds, and sounds reminiscent of ghosts and extraterrestrials are trained with artificial intelligence to explore the new auditory domain and unfold the possible connections between humans and non-humans, objects and organisms, and reality and non-reality. It intertwines machine algorithms with sound and uses the principles of an organism's behavior to expand the possibilities of self-organizing sound and the organicity of sounds in space.