Contemporary game engines such as Unreal Engine and Unity are widely used for Extended Reality (XR), yet their native audio pipelines often rely on simplified spatialization with limited acoustic control. In Augmented Reality (AR), virtual sound sources must integrate coherently with the physical environment to maintain perceptual plausibility, making accurate Six-Degrees-of-Freedom (6DoF) rendering critical. This study carried out a perceptual evaluation of multiple 6DoF audio rendering approaches, including Audiokinetic Wwise (Reflect and RoomVerb), Steam Audio, Meta XR Audio SDK, a dense 6DoF Room Impulse Response (RIR) interpolation method, and APLVirtuoso XR. A measured physical room was reconstructed in Unreal Engine 5, and the rendering pipelines were calibrated by matching the reverberation time (RT60) and direct-to-reverberant ratio (DRR) to the measured room within their respective just noticeable difference (JND) thresholds. The results showed significant differences in perceived spatial and timbral fidelity as well as plausibility and overall listening experience across the tested renderers. Despite the high accuracy achieved by the 6DoF interpolation method, an algorithmic renderer demonstrated comparable or superior performance. However, some other algorithmic renderers exhibited tradeoffs in terms of computational overhead and acoustic modelling accuracy. Our findings indicate that an optimised system prioritising a plausible auditory representation, rather than strict physical replication, may be sufficient and, in some cases, can yield superior perceptual outcomes.
