Technology Examples Test online Licensing Documentation Download Contact


The B-format for spatial audio was invented by Michael Gerzon and others in the 1970s as part of the ambisonic system of spatial sound recording and reproduction. The format consists of four channels, W, X, Y and Z, where W records the pressure at a point in space while the others record the cartesian components of the pressure gradient at the same point. The mathematical simplicity and physical significance of the B-format has ensured its survival as a topic of research, providing a steady trickle of commercial products for all these years.

Microphone systems have been designed that closely approximate the ideal, physical definition. The most common design is one comprising four separate capsules packed closely together, pointing in different directions. The outputs of these capsules can be encoded into B-format. A number of software tools are also available to create virtual sound scenes in B-format.

Soundfield microphoneDecoding

Ambisonics comprises methods for decoding B-format to loudspeaker feeds in such a way that a listener will experience an approximation of the original sound scene. However, the decoding process is not as straight-forward as the encoding process. A physical approach which reconstructs the recorded sound field is possible, but considering the volume of space occupied by even a single listener's head, it can only work for frequencies below 800 Hz in the best case. So a psychoacoustical approach must be taken instead, where the loudspeaker feeds are designed to produce a sound field which, while different from the original soundfield, reconstructs the perceived auditory scene over a larger region of space. The design of a decoder therefore requires not only a model of the wave propagation from loudspeaker to listener, but also a psychoacoustical model. As the processing power of computers increases, new methods become feasible, based on ever more complex models of the human auditory system.


Harpex is a new B-format processing algorithm that takes into account both the non-linear, parametric nature of spatial hearing and our uncanny ability to resolve simultaneous sounds. This provides a significant step upwards in quality, producing sharper, clearer and more stable renderings of B-format material than before. The following demonstration video shows some of the features of that plugin.


© 2011–2017 Harpex Ltd