A selectionist approach to auditory perception

Present signal processing methods have had difficulty in modeling ear´s ability to perceive the acoustic scene. An alternative signal processing method is proposed that extracts meaning directly from waveform features by extracting auditory information in the context of biological survival and evolution. To obtain time resolution sufficient to segregate each signal source, waveform features are converted into elementary quanta of information such that patterns in their statistical distribution can identify each source. These patterns are recognized in a hierarchy of selectionist matrices that reconstruct the spatial and temporal meanings in the acoustic environment. Some examples of computed low-level perception are shown that illustrate and support this approach