Three-party sound field sharing system based on the boundary surface control principle -Subjective assessment of voice reproduction with speaker's facing angle

A telecommunication system makes communicating more comfortable if it ensures that parties involved in distant communication feel as if they are located in the same space during their conversation. By applying physically accurate sound field reproduction, we aim to develop a telecommunication system which enables us to feel the presence of a conversational partner. In pursuit of physically accurate sound field reproduction, we have developed a sound field reproduction system based on the boundary surface control principle. We have also developed a two-party sound field sharing telecommunication system using that reproduction system. In this paper, we describe an extension of that system to three-party system and conduct the subjective assessment of its voice reproduction. In pursuit of decreasing the amount of real-time convolution calculations, we applied Gram-Schmidt orthogonalization to reduce the number of secondary sound sources. In a three-party conversation, it is important to know “who talks to whom”. Accordingly, when one of conversational partners turns towards another partner in three-party conversation, we reproduce natural changes in voice directivity caused by head rotation by detecting facing angle through image recognition and by adjusting the voice filter to suit that angle. However, this requires the voice reproduction with accuracy enough to acoustically perceive “who talks to whom”. Thus, we conducted subjective assessments of the speaker’s facing angle both in real environment and in sound reproduction environment. As a result of average angle error in sound reproduction environment, we found out that the system reproduced voice with accuracy enough to perceive who talks to whom. And we also found that there was little difference in the voice facing angle between perception in the real environment and in the sound field reproduction environment for a half of the subjects.