Real - Time Voice Transmissions over the Internet

This research identi es the problems encountered in transmitting voice over the Internet and proposes approaches to solve these problems. The current Internet is not very suitable for transmitting real-time data because its underlying protocols and switches were only engineered to transmit non-real time data. The problems posed by voice over the Internet are studied by conducting a series of experiments. These problems caused by high loss, large delay, and jitter will seriously a ect the transmission quality. A good design thus needs to combine di erent components that solve these problems together. Silence removal and compression is used to reduce bandwidth usage. Jitter bu ers are used to smooth the burstiness in the received stream caused by the network. To conceal loss, we investigate existing methods and propose two new nonredundant reconstruction algorithms based on a simple but e ective average reconstruction scheme. One is to apply adaptive ltering on top of average reconstruction in order to explore the signal trend and to obtain better estimations. E ects of important lter parameters, such as lter length and adaptation step size, are studied. Another new method, called transformation-based reconstruction method, is developed to handle low reconstruction quality caused by some rapidly changing parts of signals. Its basic idea is to let the sender transform the input voice stream, based on the reconstruction method used at the receiver, before sending the data packets. Consequently, the receiver is able to recover much better from losses of packets than it would without any knowledge of what the signals should be. This method can improve reconstruction quality without signi cant computational cost and can be easily extended to di erent interleaving factors and di erent interpolation-based reconstruction algorithms.