Abstract: Compressed audio video data is constituted by a sequence of video information blocks each of which constitutes one unit by a predetermined number of I pictures, a predetermined number of P pictures and a predetermined number of B pictures. In a case of reproducing one of the pictures of the compressed audio video data as a still picture by motion compensation prediction, when a picture stops at the B picture, one of the I pictures closest to the B picture or one of the P pictures closest to the B picture is selected, and the I picture or the P picture thus selected is displayed on a monitor in place of the B picture.