PURPOSE: To provide the voice pitch converting method capable of absorbing the error of the given peak position and constituting a high-quality synthetic voice by determining the shape of the window function for cutting out the wave-form in consideration of the input voice wave-form and the target pitch pattern for voice pitch control.

CONSTITUTION: In a wave-form cutting/overlapping section, the distances (wlf, wlb) between the peak position applied with the time window function and the front and rear peak positions adjacent to the peak position in time sequence among the local peak positions of the input voice obtained from a peak position extraction section are determined (101), these distances are multiplied by the coefficient C obtained from the conversion ratio R to the target pitch to determine the window length (103), and the right and left asymmetric wave-form cutting window functions in consideration of both the input voice and converted voice are generated (104, 105).