Aiming at the objective restriction of the current wireless mobile network, a practical video delivery solution based on the automatic generation of dubbing video slides is presented. We guarantee the continuity of the audio stream first, and then select important frames in the dynamic sampling frequency based on the visual content analysis according to the current network bandwidth. Those frames are transmitted and played in a synchronous mode with the corresponding audio track. Our solution can satisfy the user′s requirements for video content comprehension while remarkably reducing the quantity of data flow.