On improving voice quality degraded by packet loss in data networks

In voice over data networks, packet loss can have a major impact on perceived voice quality. The impact of packet loss on perceived voice quality depends on several factors, including loss pattern, codec type and packet size. Many techniques were developed to recover lost packets. In this paper a new recovery technique called "switched recovery technique", SRT, which is developed by employing the suitable recovery technique from known techniques to an estimate of the lost packet content (silence/voiced/unvoiced) depending on preceding and following packets that received correctly. The proposed technique was tested with different codecs such as LD-CELP, LPC-10, and variable rate coders. Two models of loss have been used in this study, random model and burst model to simulate the packet loss