Parametric Spatial Audio in Hearing Performance Evaluation of Hearing-Impaired and Aided Listeners

Clinical hearing performance testing has conventionally been conducted by measuring the detection thresholds for narrow-band stimuli or the reception of speech in quiet. However, especially hearing-impaired individuals and hearing device users generally have the most problems in communication situations involving dynamic background noise or reverberation, and these problems are not necessarily reflected well with these measures. Thus, methods have been developed to more accurately assess the real-life representative hearing performance using multiple loudspeakers. However, many of the current systems are either too large and complex for clinical environments or do not reproduce the spatial characteristics of the sound scene correctly. Directional Audio Coding (DirAC) is a parametric perceptually-motivated spatial sound reproduction technique. With DirAC, the spatial attributes of sound can be captured and reproduced with arbitrary loudspeaker setups. In this thesis, DirAC was applied to reproducing sound scenes and reverberation for realistic hearing performance testing. A method is proposed for conducting speech intelligibility assessments in realistic pre-recorded sound scenes where an external speech target is augmented. With listening tests, a comparison was made between a reference sound scene and a reproduced scene where the reference was reproduced by the method under investigation. Furthermore, analysis of the error sources was provided when the system is used in aided listening, where the test subject listens the reproduced sound scene through a hearing device beamformer. Requirements were specified for a clinical implementation of the system. The main advantage of the system is that with certain limitations it enables a fairly accurate assessments of real-life hearing abilities using a relatively compact loudspeaker setup.