An out-of-head localization device according to this embodiment includes an adder that calculates a common-mode signal of stereo reproduced signals, a ratio setting unit that sets a subtraction ratio for subtracting the common-mode signal, a subtraction unit that subtracts the common-mode signal from the stereo reproduced signals at the subtraction ratio and thereby generates corrected signals, a convolution calculation unit that performs convolution on the corrected signals by using spatial acoustic transfer characteristics and thereby generates a convolution calculation signal, a filter unit that performs filtering on the convolution calculation signal by using a filter and thereby generates an output signal, and headphones that output the output signal to a user.