Decentralized headphone based spatialized audio conferencing for low power devices

This paper describes a novel media distribution model for headphone based spatialized audio conferences. In contrast to existing models, our approach permits high quality spatialized audio conferencing across low power terminals, without the requirement for centralized spatial processing functionality. The model is flexible in terms of the spatial processing techniques, though we focus upon one particular implementation which employs accurate binaural room impulse response (BRIR) filtering. Our model operates by performing a single binaural filter operation upon each user´s audio at source, before encoding and transmitting the binaural signal to a centralized mixing location. Since only centralized mixing is required, this model can easily be implemented within the IP multimedia subsystem (IMS) using existing IMS components. A discussion on codec selection for this model is also offered. Finally, the impact upon network load for our proposed approach is analysed, using a traffic model based upon a hypothetical core network partition.