Holophony versus ambisonic: Deriving a hybrid method for 3-D sound reproduction in videoconferencing

Abstract

Current research in videoconferencing intends to introduce 3-D reproduction methods both for image and sound. This paper will deal only with the problem of 3-D sound reproduction. First, one specific requirement must be borne in mind: in a videoconference, the listening area has to be wide, since it is addressed to several listeners. Two approaches have been considered: Holophony—acoustical equivalent to holography—and ambisonic. In a previous paper, it has been shown that these two methods are based on the same principle, i.e., physical reconstruction of the acoustical 3-D field, which appears as the most relevant solution to obtain an extensive listening area. However, holophony and ambisonic achieve opposite compromises between system complexity (number of signals recorded, number of microphones and loudspeakers) and reconstruction performance (accuracy, size of listening area, implementation opportunities). The two approaches are compared in order to point out a hybrid method, which optimizes the result. First studies focus on sound pick-up and show that ambisonic is more efficient concerning the coding of spatial information. In this paper, this analysis will be completed, first by examining the sound reproduction step, and second by studying the method robustness toward the actual behavior of transducers.