The method adjusts pitch of voice in time domain using corelation of time-frequency domain so as to remove voice source division process. The method comprises the steps of: (a) deciding modulation ratio of original voice signal (P); (b) receiving one frame voice signal S(n) and detecting pitch of voice signal using conventional method; (c) deciding the number of zero's (L) to be inserted; (d) initializing half pitch value and compressing pitch using the conventional pitch dividing method when the number of zero's is a certain value; (e) comparing compressed pitch region with half pitch value; (f) sending voice signal (S'(n)) whose pitch is adjusted to a speaker when compressed pitch region exceeds half pitch value.