Physical Acoustical Validation of the Audio Space Lab at the Polytechnic of Turin

Hearing loss, affecting about 5% of the world's population, is the partial or complete inability to perceive sounds. Hearing aids (HAs) can help mitigate this problem. However, HAs require fine-tuning, usually done through simple tests in the lab or after in-field trials by users in daily life, involving several audiological visits to achieve a satisfactory fit. Using augmented reality and immersive audio to recreate daily auditory scenarios in the lab can expedite the fitting of HAs. However, establishing such advanced laboratories is expensive and complex. This study aims to validate a low-cost immersive audiovisual laboratory suitable for hospitals and clinics, namely the Audio Space Lab (ASL) at the Polytechnic University of Turin, Italy. The ASL is equipped with a spherical system of 16 speakers driven by 3° order ambisonics spatial audio rendering and synchronized with a Head-Mounted Display (HMD). The ASL reproduces a sound field at the center of the array, i.e., the sweet spot, along with a 360° stereoscopic visual scene. However, this study focused only on the validation of the spatial audio playback. The study addresses solely the physical validation of audio reproduction by performing dual comparisons between different conditions. The evaluation was based on objective metrics that model human binaural listening, such as Interaural Time Difference (ITD), Interaural Level Difference (ILD), and Inter-Aural Cross-Correlation (IACC). The validation was divided into two main phases: intra-ASL and inter-ASL validation. The intra-ASL validation focused on comparing different conditions recreated within the laboratory to select the best-performing configuration. While, during the inter-ASL validation, the acoustic characteristics of a real classroom were compared with those measured in the ASL when the same classroom was acoustically virtualized to verify the accuracy of the final audio reproduction. In the intra-ASL validation, the influence of the chair headrest and the HMD on the sound field was examined. The ITD and ILD analyses favored a chair without a headrest and showed that the HMD does not introduce significant changes to the sound field. Additionally, room acoustic treatment was tested under three conditions: adding absorbing panels behind specific speakers, covering a window with an absorbing curtain, and reinforcing the corners and walls adjacent to the window with panels. Analyses of the ITD and ILD indicated that the first acoustic condition provided the most noticeable improvement. Measurements were carried out to assess how the accuracy of the reproduced sound field deteriorates with increasing distance from the sweet spot, particularly at distances of 0, 10, and 20 cm. Results from ILD and ITD analyses showed poorer performance as the distance from the sweet spot increased. However, at distances within 10 cm from the sweet spot, the reconstructed sound field can still be considered as sufficiently accurate. Inter-ASL validation showed that the ASL can adequately reproduce real environments, but not perfectly. ITD and ILD analyses showed differences between the real and virtual classroom within the just noticeable difference limits, while IACC analyses showed values above those limits. In conclusion, the physical acoustical validation yielded overall satisfactory results. Future work will focus on the perceptual validation of the ASL to confirm the current results and determine whether the IACC inaccuracies are perceivable by humans.