In this study, we investigate efficient coding of multi-channel audio signals for transmission over packet networks. The techniques studied and developed as part of this research are based on redundancy coding and aim to achieve robustness with respect to packet losses. The resulting algorithm also addresses the needs of network clients with varying access bandwidths; the algorithm generates multi-layer encoded data streams which can range from basic mono to full multi-channel surround audio. Loss mitigation is achieved by applying multiple description coding technique based on the priority encoding transmission packetization scheme. The hierarchy of the transmitted data is derived from a statistical analysis of the multi-channel audio signal. Inter-channel correlations form the basis for estimating the multi-channel audio signal form the received descriptions at the decoder.
Layered multiple description coding of multi-channel audio signals
