« liberated frequencies (short demo ver.) » by Keigo Yoshida (Japan)

"liberated frequencies" redefines auditory pleasure by freeing AI from human-centric aesthetics. In advance, glitch, noise, voice, and experimental sounds were rated by a subject based on perceived pleasure. During the demo, AI learns from the most highly rated sounds and generates evolving soundscapes. The subject wears EEG sensors measuring theta waves (4–8 Hz), linked to auditory pleasure. When brain activity indicates increased pleasure, the AI disrupts it—altering pitch, tempo, and rhythm to deviate from the subject’s preferences. This creates a feedback loop that challenges the boundaries of comfort, asking whether such “liberated” sounds disturb or expand our auditory experience. This work will be presented as a demo at the IRCAM Forum Workshops Taipei 2025.

liberated frequencies - explores unprecedented soundscapes that defy our traditional auditory pleasures by "liberating" AI from the limitations of human-defined ‘pleasing'.
Before the production, our team gathered glitch, experimental, voice and noise sounds, which a subject later rated based on the pleasure they evoked. During demo, the AI continuously learns in real-time from the highest-rated sounds. Utilizing this sound data, the AI predicts and generates the subsequent auditory experiences, creating an evolving and immersive soundscape.
The subject in the soundscape wears EEG sensors that measure real-time theta waves (4-8 Hz) of brain activity.  According to Sammler et al. (2007), increased activity in this frequency band is typically associated with intensified auditory pleasure. However, in response to this heightened brain-based pleasure, the AI—continuously learning from the real-time EEG data—intentionally disrupts the experience. It transforms the generated sounds, subtly altering pitches, waveforms, tempos and syncopations, gradually diverging from the original sound patterns the subject found pleasurable.

This deliberate shift invites the viewer to explore the boundaries of discomfort, challenging the conventional auditory aesthetics inherently favored by human perception. Do these deliberately 'liberated' sounds merely traumatize the human senses, or do they open a gateway to new auditory expressions and possibilities?

github: https://github.com/keigoyoshida7/liberated-frequencies

HP: https://keigoyoshida.jp/room20.html