Primary segmentation of auditory scenes

This work addresses the problem of separation of a complex auditory scene by humans, a process which is known as auditory scene analysis. From hierarchal point of view, auditory scene analysis can be described as grouping of auditory entities of one level to form entities of higher level of representation. One of the main difficulties in describing this hierarchal construct is to describe the first level of auditory representation i.e. the elementary auditory entities. This work defines auditory elementary units and describes a computational model for the formation of these units, each unit represents an independent acoustic component which is part of the outcome of one acoustic source. As elementary units are the first level of auditory analysis and serve as a basis for further grouping and formation of meaningful mental representation of auditory sources, this work should be seen as a first stage towards a more general model of auditory scene analysis which includes modeling of grouping processes of auditory elementary units.