Research And Implementation Of Jitter Elimination In Live Video

In recent years,the real-time audio and video field has developed rapidly.As a high-quality and open-source real-time audio and video communication solution,WebRTC has been widely used.Delay,fluency,and clarity are the three key indicators of real-time audio and video systems.Due to the complex network environment,packet loss and delay jitter often occur.WebRTC designs a set of adaptive jitter buffer algorithms under the audio engine and video engine by setting a jitter buffer at the receiving end to alleviate its impact on audio and video quality.The goal of the adaptive jitter buffer algorithm is to reasonably adjust the buffer delay,thereby eliminating delay jitter,and achieving a balance between delay and fluency.However,after observing the performance of the algorithm in a live broadcast scenario and conducting an in-depth analysis of the algorithm design,we found that the adaptive jitter buffering algorithm under the audio engine is not suitable for the packet loss retransmission mechanism on mode,which leads to In the environment of poor network quality,although the sound quality has improved significantly,the buffer delay has increased significantly;the adaptive jitter buffer algorithm under the video engine is too conservative in calculating the effect of the video frame size on the buffer delay,resulting Even when the network quality is very good,sometimes there is a certain delay.Therefore,this topic analyzes the data of the Taobao live broadcast platform and obtains the network data of 33989 live broadcast channels.After building a WebRTC audio and video call test platform,the above problems are verified.Furthermore,this topic introduces a delay upper-bound prevention mechanism in the audio algorithm.In a high packet loss and weak network environment,it can maintain the improvement of sound quality and control the increase of buffer delay,achieving the difference between sound quality and buffer delay of balance.The JTB-? mechanism is introduced in the video algorithm,and the multiplicative descent factor is adaptively adjusted,which makes the buffer delay decrease reasonably and quickly,the average delay is reduced by 41.5%,and unnecessary image delay is reduced.Relying on the method proposed in this subject,the buffer delay adjustment is more reasonable,the audio and video performance is more prominent,and it is more suitable for live broadcast scenarios.It also has certain reference value for the optimization of other audio and video applications.