Perceptual evaluation of simulated and real acoustic scenes with different acoustic level of detail

The creation and simulation of ecologically valid acoustic environments is becoming an integral task in virtual acoustics. Such environments enable simulation of realistic daily-life situations under laboratory conditions in hearing research and audiology. Regarding the underlying room acoustics simulation, it is unclear which acoustic level of detail is required to capture the perceptually relevant effects. In this study, different levels of detail in the acoustic simulation of three diverse environments, a typical living room with a connected kitchen, a pub, and an underground station were examined. In each environment, an acoustic scene relevant for hearing research and audiology was tested. In the room acoustics simulation, the level of detail in the geometrical acoustics-based simulation of early reflections, the inclusion of diffusely reflected energy, and simplified rendering of late reverberation were varied, including conditions were only the direct sound was presented. Using synthetic binaural room impulse responses and dummy-head recordings from the corresponding real environments, speech intelligibility was examined with a matrix sentence test and an additional interferer in the scene. Spatial audio quality was evaluated using a pulse with a pink spectrum, a music sample, and a speech signal. Differences observed for speech intelligibility and audio quality are compared.