This paper introduces a method to transform voice based on modeling the radiated voice pulses in frequency domain. This approach tries to combine the strengths of classical time and frequency domain techniques into a single framework, providing both an independent control of each voice pulse and flexible timbre and phase modification capabilities.