Dumb question about audio delays (negative or positive)

[sebastiaaan]

Hey there,

I'm kinda embarrassed that I even have to ask this, but I legitimately can't remember when to use negative or positive delay values when muxing audio with video in mkvtoolnix.

I have 2 versions of the same film with different audio, however one of them is the remastered version so it's not exactly the same length as the older one.
What I wanna do is, mux the audio from the old version to the new remastered version.
The remastered version's length is 01:31:20.516
The old version's length is 01:31:19.306

Going frame by frame, I noticed that there's a difference of 42ms between two identical frames, so
the remastered version displays the same frame at 01:27:52.142
the old version displays the same frame at 01:27:52.100 so is 42ms behind

I did the same thing throughout different scenes and it's always 42ms. The thing is, I don't know if I should put a delay of 42ms to the audio when muxing or -42 ms - even though I know that a negative delay advances the audio and positive values delay the audio. It's just that I'm probably too tired for this but I remember I had some issues comprehending this kind of thing in the past.

What I need is just a simple answer whether I need to use a positive or negative value here. Because I was under the impression I need to use a negative value (-42ms) here.

Thanks so much in advance. I know this is a very dumb question but yeah.

[Skiller]

To make the old audio fit the remastered video, it needs to be delayed (positive delay).

[sebastiaaan]

To make the old audio fit the remastered video, it needs to be delayed (positive delay).

Hey thanks for the quick reply! So just to sum this up (and I'll probably use this as a guide in the future):

if the Video A is 500ms ahead of Video B, I need to use a positive delay to make Audio B work on Video A. (500ms)
If the Video A is 500ms behind Video B, I need to use a negative delay to make Audio B work on Video A. (-500ms)

This is by going frame-by-frame, assuming there's no other pauses, etc. between them.

Is that correct?

[Skiller]

That is correct.
I am however wary over the terms "ahead" and "behind" in this context.

If a certain frame is dislayed at
01:27:52.142 in the new version, and at
01:27:52.100 in the old version, the new version is not "ahead" – it actually takes longer from the start of the video to reach this frame. And if it takes longer, you need to positively delay the audio to make it take longer as well. Maybe that helps.

[manono]

To make the old audio fit the remastered video, it needs to be delayed (positive delay).

if the Video A is 500ms ahead of Video B, I need to use a positive delay to make Audio B work on Video A. (500ms)
If the Video A is 500ms behind Video B, I need to use a negative delay to make Audio B work on Video A. (-500ms)

No, unless I'm misunderstanding, that's not what Skiller said. "...if the Video A is 500ms ahead of Video B...". If by that you mean the lips move before you hear what is being said, or the gun fires before you hear the gunshot, then you need a negative delay. However:

The remastered version's length is 01:31:20.516
The old version's length is 01:31:19.306

And you're using the 'old version's' audio then, yes, that's a positive delay. And that's what Skiller was referring to. I think. Since it's nearly a second delay, unless the only difference is about .8 of a second at the beginning of the new version, meaning the film framecount is identical between the 2, just adjusting the delay won't work.

[manono]

Oops, double post.

[Ennio]

The remastered version's length is 01:31:20.516
The old version's length is 01:31:19.306

And you're using the 'old version's' audio then, yes, that's a positive delay. And that's what Skiller was referring to. I think. Since it's nearly a second delay, unless the only difference is about .8 of a second at the beginning of the new version, just adjusting the delay probably won't work all the way through.

I'd like to add that there is no standard relation between video runtime difference and delay appliance when cross-muxing audio. Or - video, depends on how you look at it.
As for (in this case, constant) timestamp difference for identical frames, it is not always so that the "longer" video shows a later timestamp for an identical frame. It can also be earlier, or even way later than the runtime difference. It is possible that black video is added at the beginning and/or end. Of course one could - righteously so - argue that they wouldn't be identical videos. Next to black frames, runtimes can also be affected by applying different headers/suffixes to further identical videos. You see this often when comparing same movies.
In OP's case, I'll ignore framerate difference that can be in play.