It's a very simple idea. In goes some audio and image data, and out pops a video. Specifically designed for podcasts recorded with separate tracks for each speaker.