Study On The Key Algorithm Of Real-time Audio Communication Based On The Internet

With the development of network technique and multimedia technique, themultimedia network communication technique has great progress.Especially todaywhen wireless network technique becomes more and more popular, person can usecommunication equipments which are easily carried to do video and audiocommunication any time, by wire or wireless network.When doing real-time audiocommunication by wire or wireless network,if network environment of the channelhas a big change over a short time,it makes packet loss, delay and delay jitter happen,which should make users uncomfortable.Packet loss is a key problem of influencing real-time audio quality. Usually, packetloss processing is the solution of this problem.Packet loss processing technique isdivided into packet loss recovery technique and packet loss concealmenttechnique.Packet loss recovery technique is working under the cooperation of sendingend and receiving end,it’s a little hard to implement,and it has a very good effect atprocessing packet loss.As the most common error-correcting codes-BCH coding, Turbo coding and R-Scoding, this paper is focusing on basic knowledge of R-S coding, theory of R-Scoding and decoding and applications range, based on brief introduction of BCHcoding and Turbo coding. An adaptive forward error correcting algorithm is proposedto reduce the extra bandwidth loss and network congestion to resist packet loss. Thispaper also does the simulation testing to make sure whether the algorithm is viableand right by simulate packet loss. Its every steps to improve is as follow:1）Establishing feedback between sending end and receiving end to get previouspacket loss rates calculated at receiving end;2）Calculating variance among previous packet loss rates to make the networkenvironment clear and get number of samples of previous packet loss rates which isused to predict packet loss rate;3）Get weight value of every previous packet loss rates as samples based onvariance among previous packet loss rates and number of samples of previous packetloss rates;4）The not quit accurate next time packet loss rates are predicted by doing curvefitting using two times which are closest to the next time as parameters;5）Get the exact next time packet loss rate using samples of previous packet lossrates and their weight value to update the not quit accurate next time packet loss rate.