To output a monitor voice which can be heard in high-speed dubbing.

As readout control to a buffer memory in which reproduced data are temporarily stored, read is performed by using a read address RA for performing the read to access all the addresses in the buffer memory at a read speed of N-fold. In addition, the read is made to be performed by a read speed of one-fold, and the read is performed by a read address RA2 for monitoring to access the data of divided amount of addresses corresponding to 1/N of the amount of unit addresses. Thereby, reproduced data by continuous N-fold speed can be outputted without lacking the reproduced data, and the output of reproduced data by one-fold speed is enabled although the continuity of original reproduced data is lacked.