Key points are not available for this paper at this time.
An object-based audio is a system that reproduces a highly immersive sound experience on multichannel audio systems. However, the production time for constructing acoustic metadata increases as the number of audio objects increases. In this paper, we propose a method for constructing acoustic metadata by coordinate estimation. The proposed method estimates the position of moving sound sources from video and utilizes it as the position of the audio objects, thereby reducing the production time. The subjective evaluation experiment was conducted to access the accuracy of the sound image localization constructed by the proposed method. The results of the evaluation experiment show almost all moving sound sources and sound images can be felt in the same direction.
Kato et al. (Tue,) studied this question.