When
Thematic Session 2: Software and Synthesis (Monday, 15:40)
Abstract
A sound innovation engine is in development at synth.is where approaches from evolutionary computation are being researched, for the breeding of genes that can be rendered into sounds of any duration and pitch. Initially, the primary focus will be on synthesizing novel timbres. Novelty in isolation won‘t suffice to make the sound innovation engine a useful tool for human musicking. The discovered sounds must be interesting and inspire human sonic creativity. For the tool to come up with inspirational sounds, it requires human feedback, for use as reference in the evolution process.
Currently the engine solicits feedback from humans in an interactie evolutionary compututation setting, where generations of sounds are presented and the interacting curator can select the most interesting individuals as parents for future generations. Automated evolution of sound-genes can also be directed with pre-trained classifiers, where the curation of interesting sounds is performed after potentially long batch runs. Involving human feedback directly in the automated evolution process could be useful, so the engine may come up with discoveries closer to what the ineracting human may like. This can for instance be achieved by user supplied reference sounds or learned preferences from previous reactions to suggestions from the engine. What other options could be considered in the context of such interaction? Direct reading and evaluation of biometric data obtained from responses to preceived sounds could be one, which may or may not be viable, while other forms of interaction may be more approachable?
The interaction between the sound innovation engine and human curators can be considered as a two way communication: The engine is influenced by human preferences while the sounds created can also strongly influence what directions the musicking may take.
Bio
Bj?rn is looking into methods for the search and discovery of diverse patterns. Pondering how those patterns can be applied to the generation of sound and rhythm. Considering what modes of interaction with the sounds are well suited to open up possibilities for creative discoveries.