yug is the pseudonym under which I produce beats and experimental music with machine learning.
I started by self-producing four albums which trace my transitioning from popular music to experimental music. I progressively improvised composition, vocalized lyrics, and digitised acoustic instrumentation, including singing voice, guitar, drums, bass and synthesisers, to eventually come out with entangled materials. Performance included video works made in collaboration with Adèle Pécout and custom gestural controllers made with the Grab-and-Play machine learning software.
After devoting myself to other creative projects, I returned to improvised composition, favouring spontaneous craft and lo-fi sounds across music genres. I currently produce abstract hip-hop beats in collaboration with Kévin Raharolahy (erisan) that explore the raw sounds and haunting aesthetics spawned from neural audio synthesis. I also continue to produce songs which I use as material to experiment with various generative models, including AudioCLIP, VQGAN, and more.
Performance @ EAVI XVII, Goldsmiths University of London, London, UK (April 2016)