Validation of realistic acoustic environments for listening tests using directional hearing AIDS

Assessing the real-life benefit of hearing devices inside realistic acoustic environments (RAEs) has long been desired. However, no verified methods exist in literature. This study compares the short-time SNRbenefit provided by two multi-microphone directional hearing aid strategies (fixed cardioid & binaural beamformer) operating inside: (1) a real room; (2) the real room environment generated using room acoustics modelling; (3) the real room environment reconstructed using a 4/7 Mixed-Order Ambisonics system. The modelled RAE (2), compared to the real room (1), incurs small median SNRbenefit errors (<2 dB), although the short-time errors are much larger (up to ±10 dB). The Ambisonics-coded RAE (3), compared to the modelled RAE (2), incurs smaller median errors (<1 dB), while also preserving the short-time values remarkably well. The encouraging results require experimental verification with subjective measures (e.g., speech reception threshold benefit).