A method of generating a modified video file using one or more processors is disclosed. The method comprises detecting at least one object represented in the original video file using a computer vision object detection technique. A frequency of motion of the at least one detected object is determined. A playback speed of the original video file or a playback speed of an audio track is adjusted, such that the frequency of motion of the at least one detected object corresponds to a characteristic frequency of the audio track. The audio track and the original video file are combined such that motion of the at least one detected object is synchronized with the audio track.