Modelling of the human perception of sound using ANNs

The human auditory system perceives sound in a much different manner than sound is measured by modern audio sensing systems. The most commonly referenced aspects of pure-tone auditory perception are loudness and pitch which relate to the objective measures of frequency and sound pressure levels. Current, accepted models for the perception of sound operate at a discrete level, modeling the perception at defined frequency and sound pressure level intervals and inferring how sound is perceived in non-defined regions. This paper describes an efficient and accurate method for the conversion of the sensed audio factors to perceived loudness and pitch. This method is achieved through the modeling of the auditory system using artificial neural networks. These will be shown to generate a continuous output which may have certain advantages over previous methods with regard to the ability to infer non-defined regions of perception mapping.