It is important to understand how the human brain processes speech sounds in order to lead to perception. fMRI and EEG studies have shown that certain cortical regions are activated after hearing speech, but these techniques lack the temporal and spatial precision necessary to document the unique pattern of activation evoked by each speech sound. Previous studies have shown that the primary auditory cortex (A1) pattern of activity evoked in response to speech sounds is altered by the temporal features of the sound. In this study, we show that rats’ (Rattus norvegicus) behavioral performance on consonant discrimination tasks is similar in key respects to human performance, and can be predicted from the spatiotemporal pattern of rat A1 activity when temporal information is maintained. When temporal information is ignored and the mean firing rate is used, behavioral performance cannot be predicted as accurately. We also document the ability of rats to categorize speech sounds by voicing or gender on the first day of training. This categorization ability is also predicted by the spatiotemporal pattern of A1 activity. Finally, we show that training on multiple speech discrimination tasks increases the proportion of neurons responding to low frequency tones, the threshold of A1 neurons, the response strength to tones, the receptive field size, and response latencies. Passive exposure to speech sounds increases the proportion of neurons responding to high frequency tones, and decreases the threshold of A1 neurons and response latencies. These results indicate that training on multiple speech tasks does not result in stimulus specific response enhancement in primary auditory cortex, but instead, results in generalized enhancement of untrained sounds following speech training. This result suggests that non-primary or higher cortical areas, as opposed to A1, may exhibit stimulus specificity after speech sound training.