File(s) not publicly available
Experiments With a New Customisable Interactive Evolution Framework
journal contribution
posted on 2023-06-08, 00:25 authored by Nick CollinsThis article collates results from a number of applications of interactive evolution as a sound designer's tool for exploring the parameter spaces of synthesis algorithms. Experiments consider reverberation algorithms, wavetable synthesis, synthesis of percussive sounds and an analytical solution of the stiff string. These projects share the property of being difficult to probe by trial and error sampling of the parameter space. Interactive evolution formed the guidance principle for what quickly proved a more effective search through the multitude of parameter settings. The research was supported by building an interactive genetic algorithm library in the audio programming language SuperCollider. This library provided reusable code for the user interfaces and the underlying genetic algorithm itself, whilst preserving enough generality to support the framework of each individual investigation. Whilst there is nothing new in the use of genetic algorithms in sound synthesis tasks, the experiments conducted here investigate new applications such as reverb design and an analytical stiff string model not previously encountered in the literature. Further, the focus of this work is now shifting more into algorithmic composition research, where the generative algorithms are less clear-cut than those of these experiments. Lessons learned from the deployment of interactive evolution in sound design problems are very useful as a reference for the extension of the problem set.
History
Publication status
- Published
Journal
Organised SoundISSN
13557718Publisher
Cambridge JournalsExternal DOI
Issue
3Volume
7Page range
263-273Department affiliated with
- Informatics Publications
Full text available
- No
Peer reviewed?
- Yes
Legacy Posted Date
2012-02-06Usage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC