Physiologically motivated individual loudness model for normal hearing and hearing impaired listeners

Abstract

A loudness model with a central gain is suggested to improve individualized predictions of loudness scaling data from normal hearing and hearing impaired listeners. The current approach is based on the loudness model of Pieper [(2016). J. Acoust. Soc. Am., 2896], which simulated the nonlinear inner ear mechanics as transmission-line model in a physical and physiological plausible way. Individual hearing thresholds were simulated by a cochlear gain reduction in the transmission-line model and linear attenuation (damage of inner hair cells) prior to an internal threshold. This and similar approaches of current loudness models that characterize the individual hearing loss were shown to be insufficient to account for individual loudness perception, in particular at high stimulus levels close to the uncomfortable level. An additional parameter, termed "post gain," was introduced to improve upon the previous models. The post gain parameter amplifies the signal parts above the internal threshold and can better account for individual variations in the overall steepness of loudness functions and for variations in the uncomfortable level which are independent of the hearing loss. The post gain can be interpreted as a central gain occurring at higher stages as a result of peripheral deafferentation.

Toward an Individual Binaural Loudness Model for Hearing Aid Fitting and Development

“…For HI listeners, a post gain exactly opposite to the pre attenuation counteracts the effect of the pre attenuation for high levels. This leads to the same uncomfortable level as in NH and steepening of the loudness function above the hearing threshold (see Figure 2 in Pieper et al, 2018). If the post gain is viewed as a part of the IHC loss component and depends on the pre attenuation, the current implementation of IHC loss as well as that of OHC loss are both functionally comparable to the respective components of other HI loudness models (e.g., Launer, 1995;Derleth et al, 2001;Chalupper and Fastl, 2002;Moore and Glasberg, 2004;Chen et al, 2011a).…”

“…The correction filter, attenuating low and amplifying high frequencies (see Pieper et al, 2016Pieper et al, , 2018 for details), is applied to obtain a frequency-dependent absolute threshold according to ISO 389-7 (2005).…”

“…To obtain loudness in sones S m (at time step m), a power law with the exponent B is applied to the internal binaural loudness I m and the result scaled with the factor A (Pieper et al, 2016(Pieper et al, , 2018:…”

“…Listeners were the same as inPieper et al (2018). As inPieper et al (2018), a ninth NH listener was excluded from the original dataset because the listener's uncomfortable levels were identified as outliers at three of the six test frequencies.…”

“…Good convergence of the fits was achieved if the broadband IF noise data was weighted half as high as the narrowband LNN data and the weighting of the adjacent post gain differences was increased. Weighting of LNN = 1 (x6 signals), weighting of IFN = 3, weighting of post gain differences: 1(Pieper et al, 2018: weighting of LNN = 1 (x6 signals), weighting of IFN = 0, weighting of post gain differences: 1/5).Frontiers in Psychology | www.frontiersin.org…”