The Silence of Avian Archives: A Practice-led Study of Machine Listening

In this paper, I describe a creative practice at the crossroads of machine listening and avian sounds. Machine listening is the automation of aural skills such as genre and sound event recognition through a diverse range of signal processing algorithms. A large part of these computational techniques emerged from research in speech processing and music pattern recognition during the last century. Despite being prevalent in music informatics, their scientific use in bioacoustics and artistic repurposing in the sound arts are recent phenomena. In general, machine listening has not been as widely applied to the musicking with avian vocalisations as to the human compositional domain, let alone theorised in post-human terms.

More concretely, in my practice I explore the poetic possibilities of the Fluid Corpus Manipulation Project, a set of bespoke software modules designed to compose music with machine listening instruments. I apply these techniques to a unique Western Australian Magpie digital sound archive collected by behavioural ecologists for the study of avian vocal learning. In the aesthetic MSA 2022 Themed Sessions 45 sphere, I engage with the technicity of machine listening, the meanings of silence, and the relation of the two to sensory extinction: can silence reveal what the cold ear of the machine ‘listens to’ by streaming algorithmically processed absences into the void? I argue that insofar as listening implies a subjectivity, machines do not listen, hence the need for an artistic engagement with the phenomenological gap between the interpretive capacities of human subjects and the obscure technicality of machine listening systems.