Research on Jitter Buffering Algorithm Based on E-MODEL Optimized Speech Quality

With the popularization and development of the Internet, Voice over IP (VoIP) technology is more and more widely used in the communication. However, due to the Internet network protocol of "best effort" service, it cannot provide excellent quality of service for voice transmission. The network end-to-end delay, jitter and packet loss will seriously affect the voice quality at the receiver. In order to reduce the damage of speech quality caused by delay jitter, this paper proposes a new jitter buffer algorithm based on the Atzori's algorithm, in which the E-model is used to evaluate speech quality at the user level. Then it combines the statistical distribution to the delay modeling, and improves the speech collection of Atzori's algorithm to avoid the statistical error which regards the optimal delay of the last talkspurt as the buffer delay of the current talkspurt. Finally, it finds the balance between delay and packet loss, and searches for the optimal speech quality to set the buffer size. Experiments show that the proposed algorithm would significantly improve the speech quality compared with Atzori's algorithm in wired unidirectional network and wireless bi-directional network.