Large Language Models (LLMs) are central to text-to-sound systems like text-to-speech and text-to-music, reshaping musical practices through prompt-based interaction while raising concerns about authenticity, automation, and the ethical use of artists’ work. Wilding AI is a public research-creation project bringing together artists, researchers, engineers, and students to explore speculative AI futures. In contrast to techno-solutionist uses of AI, Wilding AI treats LLMs as compositional tools rather than sound generators, integrating them into Max, Ableton Live, and spatial audio environments to control parameters such as 3D sound motion. This intervention presents the collective sound installation FERAL FREQUENCIES, which puts the system into action.
Following a year-long research-creation process culminating in a two week residency at Laboratoire formes · ondes at Université de Montréal, FERAL FREQUENCIES demonstrates the aesthetic, technical, and practical implemention of the collective’s developed capabilities in AI-driven sound spatialization. The composition traverses four key themes the collective explored: Emotional Sovereignty ; Data That Matters ; The Algorithmic Shape of Stories ; and Breaking Machines / Making Kin.
The Wilding AI Collective consists of Beth Coleman, Maurice Jones, Alexandre Saunier, Portrait XO, Daniela Huerta, Sahar Homami, Debashis Sinha, Pia Baltazar, Nao Tokui, Gadi Sassoon, Heu Hsu, and Federico Visi.
The residency and presentation of FERAL FREQUENCIES is supported by the « Laboratoire formes · ondes » at Université de Montréal.
The development of FERAL FREQUENCIES at the Society for Arts and Technology is funded by the Ministère de l'Économie, de l'Innovation et de l'Énergie, in partnership with MA Scène Nationale.
The Wilding AI project is made possible by round 14 of the Goethe-Institut International Coproduction Fund, and supported by Concordia University, MONOM Studios, 4DSOUND, and Neutone Inc.