Efficient audio compression based on the perceptual coding of spatial audio

Digital compression of audio signals is one of the technological fields that has always been tightly linked to detailed knowledge on human perception. Recent advances have extended the sources for bit-rate reduction from a predominantly masking-oriented approach towards additional exploitation of spatial perceptual irrelevancies. New techniques aim at modeling of perceptually-relevant spatial cues using a parametric approach. These parametric methods result in considerable improvements in compression efficiency for stereo and multi-channel audio coders and have consequently been adopted by standardization bodies such as MPEG (standard for audio and video compression) and 3GPP (standard for mobile transmission). Besides compression efficiency, the parametric approach has shown to be applicable to spatial processing methods as well, such as rendering of virtual auditory scenes over headphones or to provide user interactivity with audio content, such as the freedom to reposition, amplify or equalize individual objects in a complex auditory scene. In this paper, the perceptual basis and methods for parametric spatial techniques will be outlined and examples of the resulting compression and processing technologies will be given.