Interstice
2026

Installation

Dimension Varies

Multichannel Surround System

Interstice is a multichannel sound installation composed of vocal residues generated through AI timbre transfer, a machine learning process that transforms the vocal characteristics of speech. Unresolved by the transformation, these sounds occupy an indeterminate region between noise and voice, oscillating without settling into either.

Distributed intermittently throughout the space, the sounds appear without hierarchy or fixed spatial relationships. Their unstable vocal quality sustains listening at the threshold where noise begins to acquire the presence of a voice. Rather than resolving into recognizable speech, they continually oscillate between multiple perceptual interpretations.

Encountered by a single listener seated within the room, this oscillation gradually extends beyond the sounds themselves. The listener's own presence, made audible through breath, enters the same perceptual field as the synthetic vocal residues, no longer separable as "self" and "installation." Interstice exists only in this encounter: it is not a fixed sonic object waiting to be perceived, but a condition that comes into being through the presence of a single listener, one at a time.