E-model improvement for speech quality evaluation including codecs tandeming

Speech quality evaluation is one of the fundamental aspects of telecommunications which has been frequently discussed since the inception of Internet telephony. Voice services providers should ensure adequate connection quality to end users. Using speech quality evaluation, they try to identify possible issues and to prevent service quality degradation as early as when the network is being designed. This article focuses on a non-intrusive method of speech quality evaluation in accordance with ITU-T recommendation G.107, known as the E-model. Real measurements that were carried out depending on codec tandeming in the communication system and causing the packet loss in IP telephony differed from results obtained using the PESQ method. This lead the authors to carry out an experiment in which the given scenario would be measured, its results assessed and a correction function that would fine-tune the E-model for tandeming was drafted.