The term ‘‘computational audition’’ encompasses top-down and bottom-up approaches in hearing, speech, and music. As an information processing approach, the main task is to relate auditory funtion (i.e.,perceptual processes and representational structures) to structure (i.e., neural circuits). It is argued that there are two main (Gestalt) problems in computational audition: (a) the well-known constancy problems in auditory perception as exemplified in Bekesy (1969, 1974) and (b) the correspondence problem in sequential integration which is the main grouping problem in temporal pattern perception. These problems are best attacked from functional psychophysics and a neuroethological point of view in terms of intercausal nets [Wirkungsgefuege; Mittelstaedt (1969); Bischof (1995)]: Connected functional input–output relations are interpreted as causal relations. Therefore, a research strategy for auditory image formation [Yost (1992)] which may allow connecting signal and information processing approaches is proposed. This functional psychophysical approach to computational audition in connection with neuroethology in research on music perception is called systematic music theory. Computational audition may offer the possibility for a ‘‘theory of music’’ in the sense of Hermann von Helmholtz and Carl Stumpf which is related to a biologically rooted theory of hearing.