Technology Of Speech Quality Assessment In VoIP Networks

From the birth of the first telephone in1876to nowadays, communicationtechnology has been in rapid development.As abasis for communicationtechnology and the most important part of that,voice communication has made agreat developmentin recent years. VoIP service is a new generationofcommunication system.With the continuous development of IP technology, IP-based voice servicesare also increasing popularity.It is all IP-based voice servicesin the fourthgeneration mobile communication system and using IP networks to transmit voicedata. However, due to IP networks provide only best-effort service, many networkfactors such as network delay,variable delay and packet loss will have greatinfluence on the quality of VoIP voice.Therefore, developing an accurate, efficientVoIP voice quality assessment model is necessary. In this paper, mainly researchon the following aspectsFirstly, this paper introduces the application of VoIP, the practicalsignificance of the development of speech quality evaluation algorithm and thecurrent research situation.Then introduces the basic architecture of the VoIPsystem and analysis of several factors that affect VoIP voice quality, mainlydivided into the network characteristics and non-network characteristics, networkcharacteristics is the main contents of this article.In the second part of the article, we describes several current mainstreamVoIP voice quality assessment methods and focuses on the PESQ and E-Model，which are two of the most commonly used method. Then,we analyzetheadvantages and disadvantages of the two methods in theory and we validate it bysimulation data. In order to overcome the shortcomings of these two algorithms,Professor Sun proposed E-Model/PESQ hybrid algorithm,which can make up forthe deficiency of using them separately in theory. In this paper, we modified the method byjoining the network jitter in the input parameters andimprovingthe delaycomputation. Finally we propose an improved E-Model/PESQ algorithm.Finally, this paper builds network simulation platform to simulate the realnetwork environment and assess the validity of the new algorithm. Then,weanalyze the simulation data by multiple regression analysis. Using a mathematicalexpression to replace the PESQ score and the data conversion process so that wehave a model for the assessment of speech quality.