We describe a computational model of diphone perception based on salient properties of peripheral and central auditory processing. The model comprises an efferent-inspired closed-loop model of the auditory periphery connected to a template-matching neuronal circuit with a gamma rhythm at its core. We show that by exploiting auditory feedback a place/rate model of central processing is sufficient for the prediction of human performance in diphone discrimination of minimal pairs embedded in background noise – in contrast to the need for additional, temporal information when open-loop models of the periphery are used. We also demonstrate that the template-matching circuit exhibits properties, such as time-scaling insensitivity, consistent with (and desirable for) perception of spoken language.