Evidence for speech-specific brain processes has been searched for through the manipulation of formant frequencies which mediate phonetic content and which are, in evolutionary terms, relatively "new" aspects of speech. Here we used whole-head magnetoencephalography and advanced stimulus reproduction methodology to examine the contribution of the fundamental frequency F0 and its harmonic integer multiples in cortical processing. The subjects were presented with a vowel, a frequency-matched counterpart of the vowel lacking in phonetic contents, and a pure tone. The F0 of the stimuli was set at that of a typical male (i.e., 100 Hz), female (200 Hz), or infant (270 Hz) speaker. We found that speech sounds, both with and without phonetic content, elicited the N1m response in human auditory cortex at a constant latency of 120 ms, whereas pure tones matching the speech sounds in frequency, intensity, and duration gave rise to N1m responses whose latency varied between 120 and 160 ms. Thus, it seems that the fundamental frequency F0 and its harmonics determine the temporal dynamics of speech processing in human auditory cortex and that speech specificity arises out of cortical sensitivity to the complex acoustic structure determined by the human sound production apparatus.