Do sound spatialisation engines affect sound localisation abilities in normal-hearing human participants?

TY - CONF

T1 - Do sound spatialisation engines affect sound localisation abilities in normal-hearing human participants?

AU - Domniteanu, Alex Razvan

AU - Gallacher, Jane

AU - Giardini, Mario Ettore

PY - 2025/9/4

Y1 - 2025/9/4

N2 - Detecting the direction sound provenance is indispensable to clinical hearing testing. The gaming industry developed solutions for immersive spatialization of sound in virtual environments [1], which could be used to create portable localisation tests for clinical use. It is unclear whether different implementations affect people’s ability to localise sound in a virtual setting. Normal-hearing participants (n=15) were invited to six sessions. Each session, after screening through automatic Pure Tone Audiometry, participants listened to pairs of sounds through audiometry headphones and pressed a button to indicate whether the second sound came from the left or the right of the first, in a 2-alternative forced choice study [2]. Participants listened to recordings of 800Hz Pure tones virtually spatialized 1.9m away from the listener in Unreal Engine 5.3.2 (UE5) using one of the following engines: UE5 Panning, Google Resonance, Steam Audio, project Acoustics, FMOD Studio and Wwise. The angular separation between the two sounds decreased logarithmically, following the Cardiff Visual Acuity test protocol [3] and started at 10⁰. This was repeated at 6 more azimuth positions. The smallest identified angular separation was recorded at each azimuth and Bland-Altman comparison plots were generated. Few outliers indicate the measurements are similar, although bias separates the engines into three groups. The presence of bias confirms that the spatialisation engine directly impacts the localisation abilities of people and the choice of engine must be carefully considered. A comparison to real sound and wider angular separations will help inform this decision in the future. [1] K. Rogers et al., 2018, In proceedings of 2018 CHI Conference on Human Factors in Computing Systems, pp.1-13. [2] F. A. A. Kingdom et al., 2016, “Psychophysics: A Practical Introduction”, Second Edition. Academic Press [3] T. O. Adoh et al., 1994, Vision research, vol. 34, no. 4, pp. 555-560.

AB - Detecting the direction sound provenance is indispensable to clinical hearing testing. The gaming industry developed solutions for immersive spatialization of sound in virtual environments [1], which could be used to create portable localisation tests for clinical use. It is unclear whether different implementations affect people’s ability to localise sound in a virtual setting. Normal-hearing participants (n=15) were invited to six sessions. Each session, after screening through automatic Pure Tone Audiometry, participants listened to pairs of sounds through audiometry headphones and pressed a button to indicate whether the second sound came from the left or the right of the first, in a 2-alternative forced choice study [2]. Participants listened to recordings of 800Hz Pure tones virtually spatialized 1.9m away from the listener in Unreal Engine 5.3.2 (UE5) using one of the following engines: UE5 Panning, Google Resonance, Steam Audio, project Acoustics, FMOD Studio and Wwise. The angular separation between the two sounds decreased logarithmically, following the Cardiff Visual Acuity test protocol [3] and started at 10⁰. This was repeated at 6 more azimuth positions. The smallest identified angular separation was recorded at each azimuth and Bland-Altman comparison plots were generated. Few outliers indicate the measurements are similar, although bias separates the engines into three groups. The presence of bias confirms that the spatialisation engine directly impacts the localisation abilities of people and the choice of engine must be carefully considered. A comparison to real sound and wider angular separations will help inform this decision in the future. [1] K. Rogers et al., 2018, In proceedings of 2018 CHI Conference on Human Factors in Computing Systems, pp.1-13. [2] F. A. A. Kingdom et al., 2016, “Psychophysics: A Practical Introduction”, Second Edition. Academic Press [3] T. O. Adoh et al., 1994, Vision research, vol. 34, no. 4, pp. 555-560.

UR - https://biomedeng.org/biomedeng25/

M3 - Abstract

T2 - BioMedEng25

Y2 - 4 September 2025 through 5 September 2025

ER -