Comparison of virtual sound source positioning with amplitude panning and Ambisonic reproduction

Abstract

Different kinds of panning algorithms can be applied to render virtual sound sources in pantophonic or periphonic loudspeaker setups. One popular method is constant energy amplitude panning, where only the loudspeakers nearest to the virtual sound source are used for sound reproduction. Another well established reproduction method is Ambisonics, which uses the spherical harmonics theory in combination with knowledge from psychoacoustics to render virtual sound sources by simultaneously using all loudspeakers in a setup. Technical details and background information on both algorithms are given, and technical implementation issues (e.g., required computational effort) are discussed. A series of psycho- acoustic listening tests was performed to evaluate virtual sound source localization with both reproduction methods for a six-loudspeaker pantophonic and an eight-loudspeaker periphonic setup. Localization accuracy and the subjects’ reaction time were measured. The results show that the localization blur does not depend on the virtual sound source position for Ambisonics, while amplitude panning causes a high localization blue for virtual sources between the loudspeakers and a low localization blur for images close to the loudspeakers. Similar results were found in tests carried out to ascertain how dependent the reaction time is on the virtual sound source position.