Has anyone seen reports that this is also occuring on real VCDs, or is it a "File only" issue ?
This occurred on real VCDs through access = vcd. tested on win32
I'm not 100% sure, but i think it has something to do with the fact that we expect to be able to do byte accurate seeking, while in reality we can only do "sector accurate" seeking. I still haven't fully found the issue, but it's gonna be too late for 0.8.6 i fear.
I have some potential patches for trunk, which at least seem to make the issue less common, after test2 i'll see what i can do with them.
BTW. it's broken since 0.8.4, because the vcd module only started working on win32 for images, since that version. Before it was simply skipped. Note that usually the issue disappears after a pause/play, and that the file module, even though it doesn't skip, also doesnt' feature chapter/title detection.
I still have a bit of code in my private trunk, that seems to eleviate this somewhat, but doesn't fix it yet.
I'm putting this bug off to the next release again
I'm not 100% sure, but this does seem to improve the situation slightly for me. The duration calculation vs. bytesize of the "data" matches better now, causing the audio not to "drift" anymore. However 1 in 2 times, it still seems to start wrong. If anyone wants to take a look, there welcome.