AN ASTONISHING RANGE OF VOCAL SOUNDS
(2022)21:32, sonic essay, custom-trained AI-generated birdsong and music
An Astonshing Range of Vocal Sounds is an experimental sonic essay featuring creative nonfiction writing and custom-trained AI generated birdsong and music. Commissioned by Fermynwoods Contemporary Art (UK) for their series, Xylophobia, the piece surveys the use of data to study birds and introduces the problems of bias and misclassification found in avian datasets. Noting that female, intersex, and queer birds are often missing from scientific study, the work troubles what it means to be “missing” from data, as it leads to being both understudied and unsurveilled. The soundscapes in the piece were created by custom-training audio-generation AI (neural networks) on two datasets: Real birdsong data of female and non-binary presenting birds gathered from the Cornell Lab of Ornithology’s Macaulay Library, and the canon of Western classical music inspired by birdsong. Each of these datasets features male biases, and so the newly generated birdsong and music add missing birds back in without reproducing the harms of gathering real bird data.
Commissioned by: Fermynwoods Contemporary Art, Northhamptonshire, UK, 2022.
Supported by:
- CalTech-Huntington Art + Research Residency
- New Music USA Creator Development Fund
Notes:
- An Astonishing Range of Vocal Sounds is a proof-of-concept for the feature-length documentary Salvaging Birds, which features speculative bird images, birdsongs, and music alongside interviews about bird datafication and avian AI.
- With special thanks to Marie-Chantal Hamrock, Astrid Björklund, James Steventon, Jessica Harby, Sapphire Goss, and the Cornell Lab of Ornithology’s Macaulay Library.
- Comissioned image by Cassie McQuater