Abstract
Contemporary Voice-Over-IP (VoIP) systems typically ne- gotiate only one codec for the entire VoIP session life time. However, as different codecs perform differently well under certain network conditions like delay, jitter or packet loss, this can lead to a reduction of quality if those conditions change during the call. This paper makes two core contributions: First, we compare the speech quality of a set of stan- dard VoIP codecs given different network conditions. Second, we propose an adaptive end-to-end based codec switching scheme that fully conforms to the SIP standard. Our evaluation with a real-world prototype based on Linphone shows that our codec switching scheme adapts well to changing network conditions, improving overall speech quality.
Caj-Julian Schnelke
