Digital Filters for Accurately Verifying the Performance of Hearing Protectors in Use

摘要

The performance of hearing protectors can be assessed in situ by measuring the sound pressure inside the ear canal following the Microphone In Real Ear (MIRE) protocol. Thus a custom-made earplug has been designed with an inner bore, allowing the insertion of the MIRE measurement microphone. However, the actual exposure level can only be accurately predicted if the relationship, henceforth called transfer function, between the sound level at the microphone and at the eardrum is known. Previous research has revealed that the transfer function can be precisely approximated with an individualized FDTD model, but a simplified method is needed for practical implementation due to the time-consuming nature of this numerical technique. In this matter, a one-dimensional analytical model appears inadequate, hence an approximation to the detailed FDTD model based on digital filter design is proposed instead. Two different approaches have been applied to estimate the individualized filter coefficients: multiple linear regression and Multivariate Orthonormal Vector Fitting (MOVE). In general, both methods can predict an individual's transfer function quite accurately if the length of the earplug's inner bore and the length of the residual part of the ear canal behind the protector are known. However, MOVE seems more reliable for ears with a longer residual part.