Sestdiena 9. novembris plkst. 20:00 - 22:00
swrl.space • Kaļķu 24, Latvija
Nicola Privato (IS), "Mouja+"
Despite its extensive application in many artistic endeavors, generative AI in live music is still in its early stages. This is mainly due to the real-time requirements of performed music, where intuitive, immediate and reactive systems are often preferred. During the last year, I developed a setup for live performance using Neural Audio Synthesis. The system features two interfaces I designed at the Intelligent Instruments Lab, that leverage the playful and surprising interactions of magnets to control a series of AI models. The shape and interactions of these instruments are reminiscent of the Ouija boards used in spiritist sessions, and I use them to summon, explore and displace in the sonic space acoustic cryptids, conjured from the multidimensional foldings of the latent space.
Mouja+ is a further step in the evolution of my system that will be premiered at SWRL. It features a new pipeline, pairing latent navigation and timbre transfer performed on a real-time autoencoder (RAVE) with a diffusion model (Stable Audio). This system allows the generation and combination of structured sounds via textual prompts and their real-time resynthesis with my magnetic instruments, thus combining the potential of generative AI with the expressive capabilities of real-time neural synthesis.
Dario Sanfilippo (IT), "Constructing Realities"
Constructing Realities is a piece for a complex adaptive system and a performer interfaced via microphones and loudspeakers. The digital signal processing (DSP) network is provided with a sensing infrastructure in the sound domain, and an adaptation network recursively reshapes the DSP topologies, relationships and the sensing mechanisms themselves. The result is an autonomous system that responds to external and internal conditions and explores a continuous (though limited by the software data type) behavioral space that generates long-term coherence and formal developments. The performer, another agent within a larger recursive network involving microphones, loudspeakers, environment, and DSP, interacts by altering the physical topology of the network.
swrl.space • Kaļķu 24, Latvija