Perceptual Evaluation of Adaptative Higher Order Ambisonics Rendering

Higher Order Ambisonics (HOA) is a technology aiming to capture and reproduce 3D soundfields. HOA has many advantages in comparison to other technologies, but its main drawback is that the optimal reconstruction area, called sweet-spot is relatively small at low orders. To overcome this constraint, we propose to adapt the HOA rendering in real-time according to the listener’s position by computing the HOA decoding matrix for each listener’s location. Nevertheless, the HOA rendering is very sensitive to the loudspeakers’ disposition (and especially to their regularity) and to the decoding algorithm. For this reason, the adapted HOA rendering needs to be assessed.In this paper, we investigate the perceptual rendering of such an adaptation for a five order HOA system. A perceptual test was conducted considering the following factors: adaptation (with and without), listener’s position (10 cm, 30 cm and 60 cm offsets from the center of the system), geometry of the loudspeaker array (spherical and cubic), decoding algorithm (Energy preserving and All-Round decoding) and optimization (none and in-phase). Results showed that using the adaptation whatever the condition, the perception of spatial attributes of the sound source is preserved until a 60 cm translation. Moreover, without using the adaptation for the slightest translation tested here (10 cm), the perception of spatialization was less altered using the cubical geometry than the spherical geometry. Listeners did not perceive major differences between the two decoding algorithms tested here.