The Decorrelation of Audio Signals and Its Impact on Spatial Imagery

As used here, the term "decorrelation" refers to a process whereby an audio source signal is transformed into multiple output signals with waveforms that appear different from each other, but which sound the same as the source. In the experience of most sound professionals, decorrelation occurs as a by-product of other acoustic or electronic processes that often change the sound of the source. In acoustic performances, decorrelation occurs as a by-product of reverberation and chorusing, and in digital signal processing, "stereoized" reverberation and chorusing achieve the same effect. Decorrelation occurs in sound synthesis when there are slight differences between the sounds synthesized for the output channels. That often happens with granular synthesis, but can also happen with frequency modulation or additive synthesis if the composer takes special care in designing the algorithms. In the audio industry, there is a long tradition of devices for the home or studio that "stereoize" monophonic signals, and they too typically decorrelate the output channels. Numerous settings on effects processors for flanging, combing, etc. produce decorrelated output. In recording studios, vocal artists sometimes are recorded twice on separate tracks so that the micro-variations in the two performances create decorrelation. Why focus on decorrelation as a separate aspect of these processes? In the field of spatial hearing, signal decorrelation is known to have dramatic impact on the perception of sound imagery. The degree to which sounds are decorrelated has proven to be a significant predictor of perceptual effects, both in natural environments and in audio reproduction. Therefore, all of the diverse processes mentioned above are related to each other by the impact of decorrelation on the spatial imagery of the sound. While there is a considerable literature on spatial sound processing, this literature is usually concerned with one of two goals: (1) positioning sound images at a particular location in three-dimensional space, or (2) creating three-dimensional simulated environments. These goals are important, but there are obviously many other creative potentials for spatial sound processing, and other kinds of practical problems to solve. For example, decorrelation can produce sound images with the width, depth, and spaciousness typical of natural environments while circumventing the computational burden of a full environmental simulation.