A/V length problem on capture

I am capturing video using PicVideo MJPEG. When I play the captured file, everything is fine. But when I convert the file to DivX the audio slowly gets behind and is truncated at the very end of the movie. If I open the captured file in SoundForge, the length of the resulting WAV file is longer than the video file (about 0.3% fairly consistently).

Up until now I use SoundForge's time stretch feature to squash it down, but more often than not, it only gets it within a half a second of the video file (small but still significant). Then I go through and manually remove samples until it is the perfect size. This whole process ends up taking hours. Is there a way that the audio can somehow be "tethered" to the video every __ seconds or is there a capture interface that is capable of dropping audio samples when the video is ahead of it (I know you can do the opposite)?

Any help is appreciated.

[Craig Tucker]

edit

[FOO]

We have to know what capture hardware and what software.

[Riccardom]

When i followed one of the guides to transform 24fps in 25 fps (PAL land), i discovered that SoundForge, despite being my preferred audio prog, is not precise in the time warp function.
I found that GoldWave, indicated in the guide, can do a perfect job, up to the undredths.
Give it a try.

Riccardom

I have a Medion 2.66 GHz Pentium 4 using a NVidia WDM Video Capture card and Avance Logic AC97 sound card. I use VirtualDub for the capture software.

I recall trying to use GoldWave to compress time. Not only was it extremely slow, but the quality was rather poor. Have you experienced that?

[Riccardom]

I didn't pay so much attention on quality, because my primary goal was to achieve lip-sync audio.
It was a child movie and they watch it on 21'' TV.
But I don't remember it was so nasty.

Riccardom

Okay - I figured somethings out. Good information, ahead!

1) Goldwave 5.x does provide quite good quality time warp when you use either Rate or Similarity algorithm (FFT produces an atrocious echo).

2) I tried capturing using no compression on the audio (as opposed to the MP3 compression I normally use). I noticed that the dropped frame count fell to near nothing (only about 10 frames the entire movie). The result: a perfect (or close enough) audio/video sync!

So, now the question is: Is there a setting on VirtualDub (or another program all together) that compensates for dropped video frames on the audio stream?

Let me know what you all can come up with. Thanks!

[Gazorgan]

IUVCR. You can use the Audio or Video as the master to sync to the other. 30 day free trial.

MP3 capture audio has issues, PCM is better for no dropped frames, then convert as needed.