neural materials_ algorithmic sculpture

A visual symbolic language to represent the sound object in latent space. These graphical forms are generated by StyleGan2, trained on a hand drawn-to-3D digital object dataset of ancient alchemical, electrical circuit and tech-media symbols. These are the AURA MACHINE icons, now made ‘real’. Constructed from ‘tangible’ materials of glass, wood, noise and metal, these textures represent the same base materials that were originally recorded in the concrete dataset. These transmutational objects will be amplified, activated and fed back into the system for live performance.

AURA #1 STEEL

AURA #2 NEON GLASS

sound sculptural workflow_

Before deep learning there was SYMBOLIC AI (Classical AI), a system of representing knowledge (facts & rules) via a series of symbols to determine solutions. Throughout the residency I have been thinking about the importance of abstraction and symbolism to aid understanding of complexity. The many hidden layers of neural networks and the vast computation required by deep learning can be hard for humans to comprehend, we turn to symbolic systems to categorise, order and make sense of concepts such as these. The AURA MACHINE sculptures are an exercise in trying to compute the transmutation of sonic data through latent space into physical form. You can read more about the stages of this process in the research blog.

read about the sound sculptural process …

AURA #3 WOOD_COPPER_ELECTRICITY

EARTH-SENSING