On the influence of best-effort network conditions on the perceived speech quality of VoIP connections

When voice traffic is transported over a packet-based network, a number of conditions different from the ones in the traditional circuit-switched network will have an influence on the quality of the speech signal as perceived by the users. In particular, distortion of the voice signal due to delay jitter and partial loss of signal caused by packet loss will have an important impact. Therefore, in order to determine the influence of delay jitter and packet loss on the perceived quality, an objective speech quality assessment system, called DSLA, is used to predict the mean opinion score for voice connections, which are set up on a test-bed, subject to artificially introduced network conditions. A decomposition of the voice degradation is made into different parts, due to sampling, digitization, encoding/decoding, packet loss and delay jitter, respectively. It is found that delay jitter has a devastating influence on the perceived quality, when no dejittering buffer is used. However, when the received signal is dejittered, the degradation due to jitter is similar to the one caused by packet loss.