Design And Implement Of A High-quality Adaptive Voip System

Transmitting high-quality audio has become main stream in VoIP (Voice on IP) systems, as the bandwidth of Internet is increasing. The frequency of speech ranges from 20Hz to 4KHz, whereas the frequency of sound generated by all kinds of music instruments ranges from 20Hz to 20KHz. Different encoding approaches must be involved to avoid lack fidelity, which will produce larger data size in encoding speech than in encoding music signals. Low bitrate, which is suitable for encoding speech, is used in traditional VoIP systems. And high bitrate, which is suitable for encoding music signals, is used in latest VoIP systems. In on hand, using low bitrate can't provide high-quality ensurance, which is necessary in transmitting music signals, while using high bitrate suffers from wasting limited bandwidth when there are no music signals to transmit. On the other hand, most VoIP systems use C/S architecture to provide one-to-many transmitting service. This implementation brings low scalability and high economic costs.In this paper, we firstly present an adaptive algorithm to provide both quality ensurance of audio and bandwidth efficiency. The key idea of this algorithm to discriminate speech and music audio signals and decide which encoding stragety to use. In providing one-to-many transmitting service, we secondly design a new P2P (Peer To Peer) topology, two-level multi-cast tree with super nodes, to improve scalability, load balancing and real-time ensurance. Thirdly, we propose reliable transmitting algorithm based on backup path to provide clients'QoS (Quality of Service). In this algorithm, the server maintains a backup path for each client. And the audio data will be transmitted to clients directly through its backup path when the super node is unavailable.Finally, we implement a high-quality adaptive system based on the above algorithms. The actual deployment and testing prove that the system can not only decrease the bandwidth consumption while transmitting audio signals, but also meet the reliability needed in transmitting high-quality audio signals.