PURPOSE: To provide a voice information processing device in which characteristic voice outputs are easily obtained from voice inputs.

CONSTITUTION: Male voice piece waveform data EW1 are extracted from voice compressed coded data M1 against a male voice input S1 by a feature extracting circuit 55 and are stored in a piece waveform data storage circuit 74 as piece waveform data EW2. When the data are used, a recognition output R1 against a female voice input S1 is voice-synthesized using the data EW2 by a voice data synthesizing circuit 67 and a synthesized output SY is made a voice output S2 by a voice compression decoding circuit 20. Since a conventional device is used for a voice information recording and reproducing device 51, a device is easily constituted to obtain a featured voice output S2 from the voice input S1. If the above scheme is applied to a telephone equipment, the voice input S1 of a female, who picks up a handset, is made the voice output S2 by a male voice and is transmitted to a receiving side.