Comparative study of unsteady loudness models for mechanical and real sounds

Abstract

While noise levels are most often quantified using physical quantities including A-weighted sound pressure level, these metrics do not adequately represent the human perception of the noise. For this, loudness is a more appropriate acoustic metric as it describes the perceived acoustic intensity of a sound. Given that real sounds are often unsteady, a most useful loudness calculation will also account for the perceptional phenomena of time and temporal masking. Studies have been done which demonstrate the performance of several loudness models for pure tone sounds and compare these results to the ISO 226 equal loudness curves. This investigation goes beyond that and evaluates the performance of two unsteady loudness models, the Glasberg and Moore model and the DIN 45631-A1 method, using mechanical and real life sounds. Through implementation ofa jury, the differences of the two loudness calculation methods are demonstrated by plotting the perceived loudness of the sounds for two different levels compared to two different reference levels.