PURPOSE: To reduce the work load of a program incorporating by recording voice superposing discriminating signals which instruct voice information superposing in plural voice information.

CONSTITUTION: When voice superposing discriminating signals are included in the demodulated output of a demodulator 16, H level pulse signals are outputted to an OR circuit 29 from an S/H circuit 19. When voice superposing discriminating signals are included in the control data against an FM sound source 5, a PCM sound source 6 and a PSG sound source 7, H level signals are outputted to the circuit 29 from S/H circuits 22, 25 and 28. Thus, the circuit 29 outputs H level pulse signals to an AND circuit 33. When H level pulse signals from an S/H circuit 36 are taken into one of the input sides of the circuit 33, the circuit 33 outputs control signals, closes a switch 30 and the voice output level is reduced to one half. Therefore, the reduction control of the voice output level is performed by hardware.