A statistical classifier for hearing protector REAT data.
J Acoust Soc Am 2000 Nov; 108(5):2621
Hearing protector attenuation data have been demonstrated to exhibit bimodal characteristics due to poorly fitting protectors. The data have been modeled with two distributions [Murphy and Franks, J. Acoust. Soc. Am. 106, 2262 (1999)]. The bimodal distribution model was used as the basis of a statistical classifier for an individual's attenuation data to identify whether or not the data come from a poor or well fit distribution. Training classes were developed from ANSI S12.6 Method B subject fit data [Franks et al., Ear Hear. 21, 218-226 (2000)] for the EAR Classic (n=25), EAR Express (n=20) and Howard Leight MAX (n=25) earplugs. Real ear attenuation at threshold was measured under circumaural headphones for each earplug with a panel of 20 subjects. Preliminary results suggest that data from 500 or 1000 Hz yield the best ability to separate the attenuations.
Hearing-level; Hearing-loss; Hearing-acuity; Health-surveys; Age-factors; Age-groups; Information-retrieval-systems; Physical-examination; Racial-factors; Sex-factors; Audiological-testing; Audiometry
William J. Murphy, Hearing Loss Prevention Section, NIOSH, M.S., C-27, 4676 Columbia Pkwy., Cincinnati, OH 45226-1998
Journal of the Acoustical Society of America