Modeling the Joint Effects of Background Noise and Caption Text Reduction on Comprehension for Cochlear-Implant and Hard-of-Hearing Users

Abstract

This paper presents a mathematical model estimating how background noise and caption text reduction jointly affect sentence comprehension for cochlear-implant (CI) and hard-of-hearing (HH) listeners. Prior research treats these two factors separately, despite them co-occurring constantly in real listening environments (e.g., following captions in a noisy restaurant). The model fits a noise component (logistic/psychometric function, from Friesen et al. 2001) and a caption text-reduction component (quadratic function, from Burnham et al. 2008's Hard-of-Hearing More-Proficient reader subgroup) independently from published data, then proposes — via sensitivity analysis rather than direct fitting, since no dataset varies both factors together — a range of interaction effects between them. This is a preprint. The manuscript has not yet undergone peer review. Code (Python/SciPy/NumPy/Matplotlib) is publicly available at: https://github.com/Howarend/noise-caption-comprehension-math-model