一种基于AI的多路实时流媒体同步方法研究  

A Research on synchronization method of multi-channel real-time streaming media based on AI

作　　者：冯文毅 屈亮 FENG Wenyi;QU Liang(CCS Enrising Information Technology Co.,Ltd.,Chengdu 610041,China;Sichuan Education Informatization and Big Data Center,Chengdu 610047,China)

机构地区：[1]中通服创立信息科技有限责任公司,四川成都610041 [2]四川省教育信息化与大数据中心,四川成都610047

出　　处：《通信与信息技术》2022年第S02期23-27,共5页Communication & Information Technology

摘　　要：讨论并验证了一种基于AI的多路实时流媒体同步播放方法,包括:获取三路具有内容相关性的直播流;针对上述直播流传输层数据标记和同步,与此同时,优化每一路视频流从录播主机端到播放端的传输质量;在客户端优化缓存同步机制,针对底层标记后的TS文件进行同步,降低延时扰动,实现三路视频流能够在用户端精准同步播放,提升客户视频观看体验。This paper discusses and verifies an AI based multi-channel real-time streaming media synchronous playing method,including:obtaining three channels of live streams with content relevance.Aiming at the marking and synchronization of the above live streaming layer data,at the same time,optimizing the transmission quality of each video stream from the recording and broad⁃casting host end to the playback end;Optimize the cache synchronization mechanism on the client side,synchronize the TS files after the bottom marking,reduce the delay disturbance,and realize accurate synchronous playback of three-way video streams on the user side,so as to improve the video viewing experience of customers.

关 键 词：多路实时流媒体 流同步 传输层 延时扰动 

分 类 号：TN915[电子电信—通信与信息系统]