Modeling Speech Intelligibility in Hearing Impaired Listeners

Models of speech intelligibility (SI) have a long history, starting with the articulation index (AI, [17]), followed by the SI index (SI I, [18]) and the speech transmission index (STI, [7]), to only name a few. However, these models fail to accurately predict SI with nonlinearly processed noisy speech, e.g. phase jitter or spectral subtraction.

Recent studies predict SI for normal-hearing (NH) listeners based on a signal-to-noise ratio measure in the envelope domain (SNRenv), in the framework of the speech-based envelope power spectrum model (sEPSM, [20, 21]). These models have shown good agreement with measured data under a broad range of conditions, including stationary and modulated interferers, reverberation, and spectral subtraction.

Despite the advances in modeling intelligibility in NH listeners, a broadly applicable model that can predict SI in hearing-impaired (HI) listeners is not yet available. As a firrst step towards such a model, this study investigates to what extent eects of hearing impairment on SI can be modeled in the sEPSM framework.

Preliminary results show that, by only modeling the loss of audibility, the model cannot account for the higher speech reception threshold (SRT) of HI people in stationary noise compared to NH. However, this approach can, to some extent, account for the reduced ability of HI people to listen in the dips as measured by a reduced masking release (MR), where MR is dened as the SRT benefit listeners obtain in uctuating noise compared to stationary noise.

The remaining causes of reduced MR that is not accounted for by the model could be due to additional effects of hearing impairment, such as broader auditory filters or deficits in temporal fine structure.