Provided in the present application are a speech and image synchronization measurement method and apparatus, and a model training method and apparatus. The speech and image synchronization measurement method comprises: acquiring a speech segment and an image segment from a video, wherein the speech segment and the image segment have a correspondence in the video; executing any one of the following operations: converting the speech segment into a specific signal, and obtaining a speech feature of the specific signal and a visual feature of the image segment by means of a pre-trained speech and image synchronization measurement model, wherein the specific signal is unrelated to a personal feature of a speaker in the speech segment, or generating a contour image of a target person according to the image segment, and obtaining a visual feature of the contour image and a speech feature of the speech segment by means of the pre-trained speech and image synchronization measurement model, wherein the contour image is unrelated to a personal feature of the target person, or converting the speech segment into the specific signal, generating the contour image of the target person according to the image segment, and obtaining the speech feature of the specific signal and the visual feature of the contour image by means of the pre-trained speech and image synchronization measurement model; and according to the speech features and the visual features, determining whether the speech segment synchronizes with the image segment, wherein the synchronization is used for representing that a sound in the speech segment matches a motion of the target person in the image segment.