For more than two decades, Daito Manabe has worked in the fuzzy spaces where code, body, sound and image dissolve into one another. A polymath whose practice spans live performance, scientific collaboration, choreography, and software design; few artists have done more to expand the meaning of ‘interaction’ or ‘musical instrument’ can mean in the 21st century. 

Google DeepMind (formerly Magenta) has been a force in the evolution of generative music technology. Building on early experiments like Magenta (NSynth, DDSP, Magenta Studio), their latest work includes Lyria 3, a generative music AI model that provides high-fidelity sound generation and creative control over AI-assisted productions. 

In this session, the two worlds intersect and combine. Before his show at Sónar, Daito Manabe joins Jeff Chang and Yotam Mann from Google DeepMind to discuss the experience of integrating Lyria into his live set - following an encounter between Daito and the DeepMind team at Sónar+D 2025. Jeff and Yotam aren’t just developers and coders, each of them are musicians and producers in their own right, whose work at DeepMind is driven by the pursuit of never heard before sounds and new ways of making music.

Together, the three will explore the process of working with an AI. What moments are intuitive? What moments are challenging? And when does an AI move from instrument to true collaborative partner? Beyond technicalities, the conversation will also address broader themes: how to maintain creative identity when working with generative models, the emerging musical languages enabled by these systems, and the future trajectory of music in our current cultural landscape.

part of AI & Music powered by S+T+ARTS