Beaufort Scale: A data soundscape for live improvisation representing the formation of storm structures

Oliver Bown

Proceedings of the International Conference on New Interfaces for Musical Expression

Abstract

Beaufort Scale is a live improvised project for one or more performers organised around the wind. The improvisers perform storms from real historical storm data. A storm event is selected, and the data is used to drive a model of a storm structure, capturing the complex folded layers of circulating air flow that form in a full-blown hurricane. These in turn shape the parameters of an audio effects array that the performers play through. The result is a data-driven improvisational system where the performers should be thought of as playing "in the storm" rather than with it or over it, a sort of improvisational challenge involving an uncontrollable force external to the performers. The storm data is also conveyed as a series of improvised performance cues taking the Beaufort Scale as aesthetic instructions ("Smoke rises vertically", "Wind felt on face", "Devastation"). These are infused with wider historical associations from the context of the storm. Poetically, the work also posits an "ambivalence of extremes" where perfectly still and perfectly chaotic may be seen as having qualities in common, not shared by their intermediate phases. Beaufort Scale is realised through a system of agentic AI interactions in which an agent is used to create the DSP effect processor, a VST3 audio plugin, support the generation of storm reanalysis data, leading to a numeric and textual description of the storm's journey in terms of its various structural components, and then support the mapping of this storm data into instructions for the VST plugin. The plugin implements a series of grain delays operating on a ring buffer of incoming audio, then passing through a bandpass filter and distortion phase with multiple distortion options. This is a powerful parametric audio effect capable of spawning multiple sonic objects at once. It can also manage the delivery of performance cues to musicians. The plugin reads timeline data from a JSON file describing the evolution of grain streams and their parameters. The performance at NIME 2026 will use data from storm Daria which ravaged British Isles in January 1990 before continuing through Scandinavia. The lead author was 12 years old and was told in no uncertain terms they were not to go out to do their paper round, as metal bin-lids flew across the street. Other storms were brewing in 1990. The Berlin Wall had fallen, the internet was about to give birth to the Web, and house music was in its heyday.

Citation

Oliver Bown. 2026. Beaufort Scale: A data soundscape for live improvisation representing the formation of storm structures. Proceedings of the International Conference on New Interfaces for Musical Expression. DOI: 10.5281/zenodo.20782063 [PDF]

BibTeX Entry

@inproceedings{nime2026_music_12,
 abstract = {Beaufort Scale is a live improvised project for one or more performers organised around the wind. The improvisers perform storms from real historical storm data. A storm event is selected, and the data is used to drive a model of a storm structure, capturing the complex folded layers of circulating air flow that form in a full-blown hurricane. These in turn shape the parameters of an audio effects array that the performers play through. The result is a data-driven improvisational system where the performers should be thought of as playing "in the storm" rather than with it or over it, a sort of improvisational challenge involving an uncontrollable force external to the performers. The storm data is also conveyed as a series of improvised performance cues taking the Beaufort Scale as aesthetic instructions ("Smoke rises vertically", "Wind felt on face", "Devastation"). These are infused with wider historical associations from the context of the storm. Poetically, the work also posits an "ambivalence of extremes" where perfectly still and perfectly chaotic may be seen as having qualities in common, not shared by their intermediate phases. Beaufort Scale is realised through a system of agentic AI interactions in which an agent is used to create the DSP effect processor, a VST3 audio plugin, support the generation of storm reanalysis data, leading to a numeric and textual description of the storm's journey in terms of its various structural components, and then support the mapping of this storm data into instructions for the VST plugin. The plugin implements a series of grain delays operating on a ring buffer of incoming audio, then passing through a bandpass filter and distortion phase with multiple distortion options. This is a powerful parametric audio effect capable of spawning multiple sonic objects at once. It can also manage the delivery of performance cues to musicians. The plugin reads timeline data from a JSON file describing the evolution of grain streams and their parameters. The performance at NIME 2026 will use data from storm Daria which ravaged British Isles in January 1990 before continuing through Scandinavia. The lead author was 12 years old and was told in no uncertain terms they were not to go out to do their paper round, as metal bin-lids flew across the street. Other storms were brewing in 1990. The Berlin Wall had fallen, the internet was about to give birth to the Web, and house music was in its heyday.},
 address = {London, United Kingdom},
 articleno = {12},
 author = {Oliver Bown},
 booktitle = {Proceedings of the International Conference on New Interfaces for Musical Expression},
 doi = {10.5281/zenodo.20782063},
 editor = {Lia Mice and Nicole Robson and Tara Pattenden},
 issn = {2220-4806},
 month = {June},
 note = {Live Performance},
 numpages = {3},
 pages = {43--45},
 presentation-video = {https://tinyurl.com/z3rxbwt3},
 title = {Beaufort Scale: A data soundscape for live improvisation representing the formation of storm structures},
 track = {Music},
 url = {http://nime.org/proceedings/2026/nime2026_music_12.pdf},
 year = {2026}
}