Future multimedia communications systems are expected to be increasingly immersive, driven by new technology for improving 3D visual and acoustic experiences. In a classical communications sense this seems a step backwards because a much larger bandwidth is required, but the return is a richer communications mode. Acoustic technology for the audio layer is presented here, although it is not entirely inseparable from the visual aspects. The talkers' audio signals are sensed by an array of microphones assisted by a video-based talker-location system. The data is transmitted to the remote listener's location, where the 3D audio is reproduced around the listeners' ears by an array of speakers in the room. The idea is that the listener experiences being in the same room as the talkers, but achieving this convincingly has raised challenges which require new research. As a step in this direction, we use simulation in a reverberant environment to investigate the improvement from using prescribed-directivity loudspeakers for the sound field reproduction. The strong performance improvement compared to using an array of conventional omni-directional loudspeakers contributes motivation for the development of such directional loudspeakers.