Speech sounds vary continuously due to the changing configurations of the vocal tract, but the auditory system appears to be well-equipped to integrate these spectro-temporally dynamical properties in our daily speech communication. In a series of experiments it was examined how stimulus complexity and task affect perception of spectro-temporal properties associated with short (20-50 ms) and rapid vocalic transitions in speech, those which carry a great deal of information about stop consonants. By increasing the complexity of the stimulus (from tone glides to interpolated speech-based syllables) and by varying the cognitive load of the task, a continuum of conditions was created which measured both psycho-acoustical and speech perceptual abilities of normal hearing listeners. Same-different paired comparison tasks show how difference limens in endpoint frequency decrease with increasing transition duration, and increase with increasing stimulus complexity. Moreover, they show that final (VC-like) transitions are more discriminable than initial (CV-like) ones.
     If we consider our data on a perceptual continuum with stimuli and tasks increasing in complexity to match 'speech conditions' the perceptual importance of our psycho-acoustical cues seems to diminish steadily with stimulus complexity, due to partial masking effects and attentional constraints. Psycho-acoustical cues are integrated into subsequent levels of processing and the perceptual importance for all cues may decrease for complex stimuli at higher levels of processing, not only because of the limited resolving power of the auditory system, but also because the listener has to divide his or her attention over more cues. In speech this is compensated for by the combined cues and by additional stimulus properties which provide more cues for perception.