To enable a user or a lay person to catch the outline of tune before the highest or the lowest tone part and easily and surely perform key control setting by reproducing one or two part of tune before the highest or lowest tone part as a music.
MIDI data of a required music is stored provisionally in a memory part 2 according to an input from a music selection input part 3. A control unit 6 specifies a reference part to be reproduced including the highest or lowest tune part according to the MIDI data stored in the memory part 2 in response to input from a voice key control input part 5 and then performs control enabling the reproduction of the reproduced part and enabling generation of the same sound while stopping the reproduction at a time at the highest part and/or the lowest part. A voice key is determined by an input from the part 5 based on the reproduced sound, and the music is newly reproduced from the start. Accordingly a user or a person to sing can catch the outline of tune before the highest or lowest tone part thereof to sing the tune even at the highest and/or lowest tone part.