I dont think creating one file would work. I took a look at a typical DVD I use as a test case and it contains 26 VTS sets in 55 .vob files. One of these VTS sets is the main presentation and the rest are commentaries, background material, menu sets, and other assorted uninteresting stuff. If you put everything into one file, there would be no way to identify which captions are what. You could never locate the right captions for anything as there are no identifiers to link anything with. Dumping the captions from a specific .vob would easy so I have no issue with implementing that. I just find it a little ugly and dont quite understand how it would work for playback in general. Streambaby would have to process all 55 .vob files on the above DVD to play back the one set of four files containing the main program and then sort out what to do with the rest. If the program spans vobs or begins in the middle of one, it will have to work that out. When I use t2sami to feed pyTivo, it writes out both a VTS_01P2.srt and VTS_01P2.mpg pair to use to transfer VTS 1, PGC 2 programming. These contain only the program stream and synchronized captions so that only the relevant material needs to be processed. This is faster and cleaner because everything else can be ignored. My own thinking is to adapt this approach as an another alternate option in t2extract.