MP$ with AC3 into MKV: audio delay issue - help wanted.

[dumbas..]

I have a DVD 5.1 DD source audio and I want to dub this with an MP4 created with MeGUI from an HD source video.
The two source files have different 'blank' section at the front end. So, clearly they will not dub directly. (The audio of both originals has 0ms delay)
I can do this with a large amount of trial and error but..

Has anyone any advice or experience on how to dub these two files into (ultimately) .MKV without wasting too much time getting it there?

Any help would be much appreciated:thanks:

[unmei]

Is one of the audio tracks synched with the video?
If so i would synch the other one at first approximately to the synched by noting down the time of a prominent sound (loud sharp sounds like a single gun shot work very well) from playing the tracks in an audio player. This should give you at least an accuracy of 1 second, typically better because you can "guess" how much of a second elapsed even if the player doesnt show fractions of a second.

In a second step i'd open both tracks in a graphic wave editor (something like sound forge, cooledit, whatever). Since you now know about where the sound is, it is quite easy to locate the attack slope of the sharp sound in both tracks and you can determine the time difference as accurate as the difference in tracks allows. It relies on the tracks because typically two mixes will not have exactly the same waveform. If they had, you could time down to the sample (for 48khz that is about 21 microseconds). Living with the different waveforms you should still be able to be accurate enough. If i have to resynch audio, 10-20ms is usually accurate enough, even tho when i was making computer music i usually went down to ~1ms - but you just don't notice a 1ms audio desync in a movie.

With no audio reference (both tracks not in synch), determining the desynch is substantially more painful because you have to match audio to pictures, but even here sharp sounds work best as reference, esp if you can clearly assign a picture to it, such as the flash to a gunshot (unless the gun is far away, see speed of light vs speed of sound =).

Determining the offsets is all the work, writing the offset value in milliseconds into mkvmerge GUI's delay field for each track does the rest.

[foxyshadis]

Assuming the sync is correct in the original mpeg-2, you could compare it to the avisynth script from megui, and determine exactly how delayed it is.

What is the megui script, btw? It might be obvious from looking at it what the delay is, if there are trims in it.

[dumbas..]

@unmei
@foxyshadis

Thanks for the options folks

I managed to sort it out using audio delay corrector in the meantime. The delay was exactly 5 secs yes secs not ms! But it plays fine.

I timed the different intro's and managed to hit the exact delay after 3 attempts. Luck more than logic! But I will bear your advice in mind for the future:thanks: