The role of non-linear processing in the prediction of speech intelligibility of hearing impaired listeners

A speech intelligibility model is presented, based on the computational auditory signal processing and perception model (CASP; Jepsen et al., 2008). CASP employs outer- and middle-ear filtering, a non-linear auditory filterbank (DRNL, López- Poveda & Meddis, 2001), adaptation loops, and a modulation filterbank, and has previously been shown to successfully account for psychoacoustic data of normal-hearing (NH) and hearing-impaired (HI) listeners in conditions of, e.g., spectral masking, amplitude-modulation detection, and forward masking (Jepsen et al., 2008; Jepsen and Dau, 2011).

This study shows the predictive power of the speech-based extension of CASP for NH listeners in conditions of additive noise, phase jitter, spectral subtraction and ideal binary mask processing. Furthermore, the model was adapted to individual hearing loss profiles in order to study the role of different processing stages, namely the auditory filter and inner hair cell transduction, in predicting accurate speech reception thresholds of HI listeners in conditions of speech in noise. The proposed model framework sheds light into the importance of non-linear processes in the auditory system for speech understanding, and how hearing losses that may linearize auditory processing affect HI speech intelligibility.