Contribution
Assessing the Benefit of Binaural Speech Enhancement Techniques
* Presenting author
Abstract:
Recently, two binaural beamformer strategies that enhance the head shadow effect were proposed to improve speech intelligibility for people with hearing loss (Dieudonné and Francart 2018; Lopez-Poveda et al. 2022). Dieudonné and Francart’s (2018) head shadow enhancer (HSE) operates only at low frequencies. In contrast, the contralateral sound field attenuation (CSA) by Lopez-Poveda et al. (2022) generalizes the HSE and extends it to a broader frequency range. While both approaches show promise in simulated free-field experiments with research implementations of the algorithms, they have not been evaluated in a true free-field environment using commercially available hearing aids and individuals with hearing loss (HL). This study aims to fill this gap by evaluating these strategies in a traditional speech intelligibility task, and under more complex conditions involving multiple maskers and uncertain target speaker direction. Based on signal-to-noise ratio simulations, we hypothesize that the CSA offers greater benefits for individuals with severe-to-profound HL than for those with moderate HL in the traditional task, while the HSE only benefits those with severe-to-profound HL. In the complex task, individuals with moderate HL are not expected to benefit from either strategy, while the benefit for those with severe-to-profound HL depends on the beamformer’s steering parameter.