Anna Xambó
Anna Xambó is an experimental electronic music producer and researcher. Biased for being passionate about extreme digital minimalism and a past performer on bass guitar, she likes to explore the boundaries of digital sound, focusing on low frequencies, compulsive rhythms, and noisy textures. Her research focuses on sound and music computing systems, with particular attention to novel approaches to collaborative, participatory, and live-coding practices. To date, she has released four solo albums: init (Carpal Tunnel, 2010), On the Go (Carpal Tunnel, 2013), H2RI (Pan y Rosas, 2018), and detuning a tuning (Carpal Tunnel, 2023). She is currently Senior Lecturer in Sound and Music Computing at the Centre for Digital Music (C4DM), School of Electronic Engineering and Computer Science, Queen Mary University of London. There, she leads the Computational Sonic Arts Lab, co-leads the EECS Women in Higher Education Network (WHEN), and serves as technical co-director of the C4DM Studios. She is Principal Investigator on the AHRC-funded project “Sensing the Forest: Let the Forest Speak using the Internet of Things, Acoustic Ecology and Creative AI” (2023-2026) and previously led the EPSRC Human-Data Interaction Network Plus-funded project “MIRLCAuto: A Virtual Agent for Music Information Retrieval in Live Coding” (2020-2021).
Inventions for Machines: Live-Coded Dreams
Abstract
What can live coding, combined with machine listening and creative AI, reveal about situated practices of sound, place, and memory? Inspired by Delia Derbyshire and Barry Bermange’s Inventions for Radio: The Dreams, this improvised sound-collage unfolds as an open-ended exploration of live soundscaping. Drawing on archival recordings and site-specific environmental sounds from the Berlin Museum of Musical Instruments (Musikinstrumenten-Museum), the performance creates evolving sonic dialogues between past and present, human craft and machine perception. At a moment when AI’s role in musical practice is widely debated, this performance-keynote investigates environment, place, and embodied knowledge as vital human dimensions of creativity—ones algorithms may reflect, adapt to, or learn from, but cannot fully inhabit.
Mat Dryhurst
Mat Dryhurst is an artist renowned for his pioneering work in machine learning and governance. He runs Herndon Dryhurst Studio with his partner Holly Herndon, with whom he has co-produced critically acclaimed albums including Platform (4AD/RVNG 2015) and PROTO (4AD/ 2019), and developed Holly+, a protocol for shared ownership and licensing of an artist's identity and intellectual property.
In 2022 they co-founded Spawning, an organization that attempted to build a web-scale data permissions protocol for AI. Its tools Have I Been Trained and the Do Not Train standard proved influential in machine learning policy discussion.
Dryhurst's practice operates at the intersection of art, technology, and infrastructure. He describes their work as Protocol Art: work that intervenes at the level of systems and protocols rather than objects, arguing for agency at the infrastructure layer.
His work has been exhibited internationally at major institutions including the Serpentine Galleries, London; the Whitney Biennial, New York; and KW Institute for Contemporary Art, Berlin. In 2024 they published "All Media is Training Data" (König Books), a retrospective of their career and a position statement on the future of AI and art. In 2026, he presents Strange Rules at Palazzo Diedo in Venice, a major exhibition co-conceived with Holly Herndon and Hans Ulrich Obrist, presented by Berggruen Arts & Culture.
Dryhurst is a Senior Research Associate at HAI Lab and fellow at the Institute for Ethics in AI at Oxford University.He has held faculty positions at NYU and the European Graduate School. He is a widely consulted voice on AI policy, copyright, and the political economy of cultural production, contributing to debates around data rights and the governance of generative AI.
Protocol Art
Abstract
Some of the most compelling art today lies in shaping the conditions of a medium: building the models, software, and rules that govern participation, rather than simply making work within them. This talk reflects on a decade of our projects pulling at this thread.