I have had continuing problems with sound synch. Problem has been after encoding to SVCD Mpeg2 the sound runs out of synch and then it corrects itself every twenty seconds or so. I followed every possible method of framerate converting, demuxing using all programmes such as CCE, TMPGENC, VDUB. Still same problem.
I then read on the TMPGENC site that despite stretching and converting audio the programme can actually read time and date stamp data and pull the film back into line hence causing the problem i have as it will drift out again.
Now in another guide i read that if i seperate the audio / video in TMPGENC to create a M2V file, and use elementary video stream, it will create a file with no audio data. System Video still creates a video M2V file but it maintains some audio data in the video stream. I currently use system stream which Cooooooooold explain my problem. My thoughts are if ES stream has no audio data then it cannot look at time stamps etc to allow it to correct itself. My question is really what is elementary stream video and what are the differences Vs video stream? Is it really only the audio data or are there other vast differences.
+ Reply to Thread
Results 1 to 2 of 2
A program stream will put PTS (program time stamps) in the GOP header.
ES, elementary stream, create just video, and audio files. The time stamps are present in the video, but mpeg audio, nor PCM audio allow time stamps. This makes it harder for sync. The muxer will take the first frame of audio and match it to the first frame of video, no other syncing is done.
AC3 allows time stamps. If you embed the time stamps into the ac3, most good muxers can sync on a frame by frame basis.
You can try Womble's MPGVCR, it has a GOP tool that attempts to correct GOP time codes, and audio PTS errors for mpeg streams.