Interpolating image topic for synchronization with audio

edit

I am trying to do audio-visual processing, where I train the models using some samples of audio and the corresponding video frame. I recorded some audio and video data as a rosbag. Since the video framerate is approximate and not exactly 30fps, when I try to do one-to-one correspondence after extracting the images with audio, there is a drift after some minutes.

Is there a way in which I can interpolate the image topic in the rosbag with missing frames to have exactly 30fps.