Subjective evaluation of conversational multimedia quality in IP networks

Human perception of Voice- and Video-over-IP connection quality is influenced by a lot of factors, first of all the network conditions like packet loss rate, delay and jitter but also others like the hard- and software (speech or video codec) in use. In this paper an experimental setup is developed which allows adjustment of certain network parameters to study their impact on the perceived conversational quality. Furthermore, the new LEGO^reg model conversation test scenario is proposed that motivates two test persons to engage in highly interactive conversations with balanced use of both the audio and video modalities. The setup was used in a subjective assessment of different delay times and packet loss rates. The results of the assessment, which involved 28 test persons, are presented in terms of Mean Opinion Scores (MOS).