Mechanized Cacophonies

Mechanized Cacophonies is a short experimental online work that examines how experiences of nature are mediated through technological systems. AI-generated imagery of coastlines is interwoven with fragments of real footage of cypress trees, which appear intermittently and without pattern. These visuals are accompanied by a specially composed soundtrack generated using machine-learning models trained on audio collected by Caroline Sinders and Anna Ridler. The work emerged from the artists' experiences during lockdown, when restrictions significantly shaped both movement and method. Their practices in art and machine learning are grounded in the construction of bespoke datasets, and during the summer of the pandemic this process was constrained to what was immediately accessible. As a result, both artists produced audio and visual datasets drawn from their local coastlines. When used as training material for a GAN, this limited and geographically specific data produced imagery that is deliberately unstable and unreal. The tension between generated and documentary material—between the synthetic and the recorded—extends into the layered, orchestral soundscape, which viewers experience within their own domestic spaces.