Research On Key Technologies Of Real-time Streaming Of Audio And Video In Intelligent Recording And Broadcasting System

With the development of computer technology and streaming media technology,its combination with education has spawned a new generation of intelligent recording and broadcasting system.The intelligent recording and broadcasting system integrates recording,transmission and playback,which is of great significance for improving the utilization of quality education resources and alleviating the uneven distribution of educational resources.This paper mainly studies the transmission of audio and video real-time streams in the intelligent recording and broadcasting system.This paper first introduces the real-time streaming process based on RTMP,focusing on the audio and video coding technology of H.264 and AAC,the audio and video packaging technology based on FLV format,the basic working principle and forwarding technology of RTMP server.Finally,the network delay problem with the RTMP protocol is analyzed in depth,and its key problems are identified.Aiming at the network delay problem of RTMP,this paper innovatively proposes UDT_RTMP protocol for data transmission using UDT,and customizes the audio and video data encapsulation format based on UDT RTMP protocol.At the same time,in order to make full use of the transmission performance of UDT_RTMP,the receiving end of RTMP server is redefined.In order to verify the reliability of UDT_RTMP data transmission,the frame loss statistics are carried out in this paper,the statistical results show that the UDT RTMP data transmission is reliable,the maximum frame loss rate is less than 0.67%,and the average frame loss rate is 0.272%.In order to verify its real-time performance,The UDT RTMP and RTMP were compared,the experimental results show that UDT_RTMP reduces the network delay from about 2000ms to about 800ms.In order to guarantee the data transmission quality of UDT_RTMP,this paper proposes a series of network optimization strategies.Since RTT can characterize the current network congestion and its trend in the short term can be predicted,this paper innovatively proposes the RTT trend prediction based on bp neural network and the dynamic adjustment strategy of video sender.With the combination of RTT trend prediction and the suitable RTT judgment basis,the network congestion status can be predicted at least 100-200ms in advance,and the video transmission can be dynamically adjusted timely.In order to avoid the picture blur caused by video frame loss,this paper proposes the H.264 video frame loss strategy based on the UDT retransmission feature;In order to improve the smooth performance of video,this paper proposes a dynamic adjustment strategy based on ring buffer.Finally,this paper builds an intelligent recording and broadcasting system with Hi3531A,and successfully applies the system to the live broadcast of the New Year's party of a certain school.At the same time,the real-time and fluency of UDT RTMP is fully analyzed on this platform,the result shows that UDT_RTMP can realize the smooth transmission of streaming media with lower delay.