Just Spatial: Real-Time Dichotic Roughness Minimization for Stereo Music
Year: 2026
Type: preprint
Abstract: Twelve-tone equal temperament (TET), the standard tuning system in Western music, trades tuning purity for unlimited key flexibility, leaving listeners with a residual sensory roughness that can make long listening sessions more tiring. Existing approaches tried to reduce this roughness by changing the frequencies themselves (retuning, shifting, or attenuating partials), which alters the pitch and timbre of the music. This paper presents Just Spatial, a real-time system that instead leaves every frequency untouched and redistributes clashing spectral components between the two ears. Given that sensory roughness is a peripheral effect, two close frequencies beating within the same cochlear critical band, routing the conflicting components dichotically (one to each ear) avoids the beating effect and lowers the perceived roughness without removing any spectral content. Built in Max/MSP, the system computes a per-bin roughness estimate using a Vassilakis-type model on an ERB critical-band scale, then applies an equal-power per-bin pan that pushes high-roughness bins toward opposite ears. On preliminary internal measurements the system's own roughness metric shows a clear drop, strongest at the highest setting. I frame this algorithm as a spectrum-preserving, headphone-based alternative to retuning, and outline the perceptual and physiological evaluation that remains to be done.
Author: Wang, Yun-An
Institution: University of Jyväskylä
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7166332590
Cites:
Cited by:
Related to: