divx and extractstream...so close and yet so far...

[oddmanout67]

I was just curious if anyone else is needing to use a rather large 
offset for audio in order to get reasonable sync (I usually end up 
with -600 to -400 ms). The offset values I get from splitting raw ty 
files are only very rarely even close. I've used extractstream and 
splitstream for a long time, but have recently started trying 
mfs_stream and vsplit. The first time using mfs_stream/vsplit was 
pretty much right on, but since then it's way off each time, just 
like extractstream/splitstream was. 

I've been converting everything to divx using this method: First, I 
extract the raw ty file and split it. The audio I load into winamp 
and save out as mp3, which I then open in mp3gain to normalize the 
volume. The video I compress using Gordian Knot/Nandub. Once the 
video and audio are finished, I load both into Nandub and check the 
frame rate option. I compare the current frame rate and what the 
frame rate would need to be adjusted to in order for the audio and 
video to match. I divide the total number of video frames by each of 
the two rates to estimate the length of both audio and video. I take 
the difference and use that to offset the start of the audio. The 
values I get are usually between -400 and -600 ms (unless there's 
something wrong with one of the files). I know this is kind of a 
strange way to go about it, but it's the only way I can get my audio 
even remotely close to being in sync. 

My question is...the offset values I'm getting seem to be quite a bit 
different than everyone else, and if I could just use the offset 
values from splitting the ty file, it would save a few steps in this 
process. 

How can I improve what I'm doing? 

Thanks!