For speech recognition applications the ultimate goal is to achieve human-like performance. State of the art signal processing algorithms for speech recognition fall considerably short of this goal. I have attempted to reverse engineer peripheral regions of the mammalian auditory system to gain some insight into how our brains might process speech. I constructed a biologically based temporal signal processing model of the auditory periphery consisting of the following sub-models (a) generic spiking neuron; (b) middle ear; (c) cochlea; (d) inner hair cell; (e) spiral ganglion cell; (f) bushy cell; (g) stellate cell; (h) octopus cell; (1) medial superior olive multipolar cell. Individually, the models produced behaviours similar to their biological counterparts. As a whole, the model was able to demonstrate how the missing fundamental of a complex Signal could be perceived, despite its absence from the spectrum. My work has led me to conclude that (a) the peripheral auditory system appears to operate primarily as a wideband temporal processor; (b) the dorsal cochlear nucleus appears to be involved in compensating for pinna position; (c) the connectivity of octopus cells is entirely consistent with their suspected role as onset detectors; (d) the ventral cochlear nucleus appears to have specialised time, intensity, and startle output pathways; (e) the laminar architecture of auditory nuclei is likely to play an important functional role; (f) the spatial cross correlation theory of pitch perception [Loeb, White and Merzenich, Biol. Cyber. 47, 149-163 (1983)] may need further extension to account for the phase independence of pitch.