Abstract
This seminar presents the signal processing procedures that appear in the cochlea and in the first auditory areas of the cortical system. This system takes a one-dimensional sound wave as input and transforms it at various levels into representations that extract sound attributes such as timbre or pitch . The first transformations performed by the cochlea have been found to resemble the wavelet transform, which extracts features useful for many applications in audio and speech processing. At the cortical level, the cochlear representation is again transformed by two-dimensional wavelets, producing a representation with very rich attributes. These attributes are used by the brain to classify sounds, to separate complex mixtures, to achieve robust listening to different sound disturbances or to perform cognitive functions on speech, music or environmentally generated signals, in order to create categories, extract meaning or make decisions.
