A representation of the speech sound at that time. For example:
These exist as there are limitations on the rate of speech production, (thus the fundamental information transfer rate) and this is less than a general signal at the same sampling rate. Speech analysis deals with time-scales of around 20ms.