Acoustic Comfort Index (ACI): A Transparent Multi-Parameter Model for Indoor Acoustic Assessment

Abstract

 Current indoor acoustic models mainly use sound pressure level cutoffs or occupant surveys, missing an interpretable, multi-factor index that links objective metrics with how spaces actually feel. This paper introduces a fuzzy-logic Acoustic Comfort Index (ACI) that integrates six inputs: sound pressure level, dominant frequency, source type, masking condition, acoustic criticality, and noise complexity, into a continuous 0–1 score (lower indicating greater comfort) with categorical interpretation. A 2,025-rule Mamdani system is generated through automated rule construction. Validation includes 1,000 randomized cases and 16 representative indoor scenarios. Local sensitivity analysis quantifies variable influence, while a ridge-regularized GAM surrogate confirms expected monotonic behavior and yields strong hold-out agreement. PCA clustering identifies comfort regimes that support practical diagnostic use. Benchmarking against SPL-only and reduced models demonstrates the added value of contextual and perceptual variables. A MATLAB ACI Calculator App enables practical application from direct inputs or audio recordings.

Keywords: Soundscape, Indoor Environmental Quality, Acoustic Assessment, Acoustic Comfort Fuzzy Logic