Developments in communications technology, such as audio–visual conferencing and telepresence, have created a new set of demands for audio delivery systems. In the past, a telephone handset, with its limited transducers, was adequate for simple spoken communication; however, future multiuser communication systems and immersive machine interfaces will be greatly enhanced by using high-quality three-dimensional audio. Such three-dimensional sound systems can position individual talkers’ voices in space around the user, creating a greater feeling of immersion and improving intelligibility. This presentation discusses the design and implementation of a number of spatial audio system implementations. The applications range from a SmartSpace chair (a single-user environment using a transaural system) through to a VisionDome (a large multiuser environment using a combination of ambisonics and nonlinear panning). Theoretical design and practical implementation issues are discussed, including problems encountered and how they were overcome.