Sound is transformed by the early auditory system into several intermediate representations that are then projected to the higher auditory centers of the thalamus and cortex. From a signal processing point of view, the central stages extract perceptually significant features by: (C.1) multiscale analysis of the spectrum; (C.2) selective amplification of important representational features; (C.3) synthesis of cues from multiple sensory sources into unitary percepts. While the physiological details of all these representations remain uncertain, enough is known to allow us to explore the functional and computational significance of these processes as we propose in the next three subsections.