Representation of frequency variations in time in speech signals