gms | German Medical Science

Evaluation of listening performance in realistic situations or in the field is an important step in the development of advanced hearing aid processing strategies and their adaptation to individual subjects. The Portable Hearing Laboratory (PHL) was developed for such an integrated approach. It employs ear-level hardware that resembles real hearing devices in terms of handling and acoustical properties and uses the open Master Hearing Aid [1] (openMHA) software for low-latency real-time audio signal processing, which allows flexible setups of hearing aid processing chains, algorithm development and interfacing in experimental setups. In this study, 20 hearing-impaired subjects were provided with the PHL to investigate individual effects of spatial signal enhancement on speech reception in spatially complex acoustic environments. Individualized earmolds for the acoustical coupling were provided for each participant. Gains were adjusted according to the loudness-based fitting method trueLOUDNESS [2] to restore the binaural broadband loudness perception. Different spatial filtering methods that are available with openMHA were used: binaural coherence filtering, bilateral adaptive differential microphones (ADM), and binaural minimum variance distortionless response (MVDR) beamforming. For a systematic investigation of the effect of these different processing schemes on speech reception, listening situations with varying acoustic complexity in a realistic cafeteria environment were simulated with the Toolbox for acoustic scene creation and rendering [3] (TASCAR). The playback of the stimuli was conducted with a multi-loudspeaker setup using higher-order ambisonics. Speech intelligibility was assessed using a sentence test. Furthermore, a screening of listening effort on a categorial scale was conducted. The results indicate that the performance achieved with the PHL setup is comparable with the subjects commercial devices in terms of benefit in speech intelligibility as well as listening effort compared to the unaided case. Measured in terms of median words correct, the benefit of spatial signal processing compared to the condition with only amplification was 0.3% for the binaural coherence filtering, 5.0% for the bilateral ADMs, and 5.5% for the binaural MVDR beamforming. However, large individual differences between -9% and +22% for all three methods were observed and suggest that an individual adaptation of spatial enhancement may be needed.