Hearing protector attenuation: models of attenuation distributions.

Current hearing protector rating standards estimate the protection performance for a given frequency as the mean attenuation minus a multiple of the standard deviation. Distributions of real-ear attenuation at threshold data are fit with maximum likelihood estimation procedures using both normal and mixed-normal models. Attenuations from six hearing protectors, Bilsom UF-1 earmuff, Bilsom Quietzone, E.A.R Classic, E.A.R EXPRESS Pod Plugs, Howard Leight MAX, and Wilson EP100 earplugs, measured with a subject-fit protocol are reported. The mixed-normal (bimodal) model provides a better fit to the empirical data than the unimodal model for most frequencies and protectors. Primarily, the bimodal model fits the shape of the distributions caused by data from poorly-fit protectors. This paper presents an alternative method for estimating the protection performance either with the more accurate bimodal model or directly from the empirical cumulative distributions of the attenuation data.