Sound field sharing system based on boundary surface control principle

摘要

The current telecommunication systems, e.g. the telephone, transmit only a 1-channel audio signal; therefore, it could not transmit spatial information. However, if the system transmits accurate spatial information, users can communicate with each other as if they were in the same place. On the other hand, we could reconstruct the 3-D sound space accurately by using a boundary surface control (BoSC) based sound reproduction system. In this paper, first, we describe the "sound field sharing (SFS)" system in which multiple BoSC-based sound reproduction systems are connected through the Internet. The SFS system consists of one server and two or more clients. Using the SFS system, users can perceive the other speaker's facing angle and feel "in-the-same-room" presence during their conversation. To evaluate the performance of the SFS system, subjective assessments were carried out. The results show that subjects can distinguish the facing angle of their conversational partners. Next, we describe an approach to reduce the numbers of loudspeakers and microphones. The BoSC-based sound reproduction system we have constructed consists of 62 full-range loudspeakers, 8 subwoofers and 70 microphones. The system requires huge computational complexity and transmission bandwidth. We reduced the numbers of loudspeakers and microphones by using the Gram-Schmidt orthogonalization and evaluated the degradation caused by this reduction. The results of the sound image localization tests in the horizontal plane show that there is no statistically significant difference between the BoSC-based sound reproduction system consisting of 62 loudspeakers and the one with 24 loudspeakers.