Intermedia synchronization for videoconference over IP

摘要

This paper deals with the problem of audio and video synchronization issues for real-time audio-visual communication over IP-based networks. Starting from the real-time transport protocol (RTP) specifications (Schulzrinne, 1995), it provides an accurate description on how to recover a reliable absolute time reference for audio and video signals from header information in RTP and RTP control protocol packets. Such temporal informations allow to synchronize both media within acceptable perceptual bounds for reconstruction at any receiver end, in a possibly multi-point videoconference. This may occur independently of the fact that all (audio/video) packets reach destination, or that multiple replications of such packets arrive at destination. The proposed solution does not take into account the possible different delays that may be introduced between the different communication points, thus resulting eventually in potential jitters in the reconstruction of the audio-visual streams between different receivers. Each receiver handles its reconstruction independently of any transmission/processing delay. In order to ensure a better quality of the reconstructed material, priority is given to audio information. If the audio stream anticipate the video stream, the receiver simply discards video packets. Conversely, when video is ahead of audio information, the video decoding stage is interrupted till audio information arrives. Experimental simulations over a LAN have demonstrated the validity of the proposed approach.