A main speaker decision for a distributed telepresence system

In this paper, we propose a method to detect a main speaker and automatically change into one's high definition (HD) video for a distributed telepresence system, so that the users feel immersive and convenient. In contrast to centralized systems, user equipment (UE) performs the main speaker decision (MSD) with a time synchronization using network time protocol (NTP). The MSD method includes a voice activity detection (VAD) and post-corrections to remove unwanted voice detections and share the same main speaker. We emphasize an audio signal of the main speaker to become more immersive. The proposed approach is applied to the telepresence system developed by ETRI and shows good performances.