Provided is an image information summary device for reproducing the summary information, by which an operator is capable of grasping the image information appropriately by the use of the preliminary announcement information. A summary reproducing apparatus obtains the continuous voice image information which is provided in plural times and extracts a characteristic amount of each piece of voice image information. Simultaneously, upon carrying out the summary reproduction, the summary reproducing apparatus compares the object voice image information to be an object of the summary reproduction with the non-object voice image information having the preliminary announcement information in the object voice image information, detects the same shot from the object voice image information and decides the partial image information to be extracted on the basis of the same shot in this object voice image information.