Effect of ambisonic order on sound localization in the horizontal plane

Sound localization performance was measured for ambisonic panning over a circular array of loudspeakers with various orders of ambisonics. Absolute sound localization was measured by recording physical orientation to the perceived sound source using a head tracker. The frequency of front/back errors was elevated using first-order ambisonics. After front/back correction, first- and second-order ambisonics still produced distortions in perceptual space compared to higher orders. For orders three to eight, the further performance improvements were not statistically significant; accuracy was comparable with that for point sources and replicated the perceptual biases listeners have shown for the azimuth of sound sources. Precision of absolute sound localization was also poorer for first- and second-order ambisonics. To estimate precision without contamination from motor responses, the minimum audible angle (MAA) was measured at a range of azimuths for 1st-, 3rd-, and 8th-order ambisonics. MAAs were smaller for 3rd- and 8th-order ambisonics than for 1st-order ambisonics. At the higher orders, the MAA was comparable to that observed with point sources in previous studies when the source was close to the midline. Otherwise, ambisonic MAAs were poorer than reported for dicrete sound sources, but replicated reports of relatively elevated MAAs in the rear hemifield.