E Model Is Based On The Delay Jitter Expansion

With the technique of Internet and voice codec developing so rapidly,with the bandwidth improving so great,the number of VoIP user is rising in a high speed in China.Contrast to PSTN,VoIP has many advantages,such as:more proficient and cheaper.But VoIP has its own disadvantage:the quality of voice have no quality guarantee.That is why people focus on VoIP QoS.There are two kinds of method to evaluate VoIP quality:objective and subjective. Subjective method is also called MOS(Mean Opinion Score).The value of MOS is judged by trained people those who listened to the same voice under the same environment.Objective methods include PESQ(perceptual evaluation of speech quality) and E-model.PESQ divides both the reference signal and the degraded signal into smaller blocks,computes every block's Fourier Transform coefficient,gets a score by comparing these two coefficients,and then maps it to MOS.Both MOS and PESQ are off-line methods,they can't reflect delay,jitter,or loss's affect on VoIP,so they can not be used to monitor the VoIP service quality.The other objective method,E-model,is more attractive for real-time speech quality estimation as its results is based on network parameters.It is one of the most popular real-time speech quality evaluating methods.But E-model has a defect:it doesn't take account jitter.Today,lots of experts have done lots of studies on E-model, and these work can divide into two directions:one is the application of extended E-model under special environments;the other is introduce jitter into E-model.But most of there studies are focused on either delay's self-similarity or the size of jitter buffer;To compute self-similarity requires huge number of packets,that leads to low quality of real time,and the size of client's jitter buffer is unknown for monitor tool.This dissertation summarizes the above methods and proposes a new method to compute jitter's affect on the quality of VoIP.This new method analysis the affect of loss and delay variation caused by jitter on the speech quality of VoIP,and combined these affects as jitter factor;then introduces it into E-model and proves it by experiments.