Auto Gordian Knot gives delayed audio in output files

[loninappleton]

Just recently I've been having problems with audio synchronization using Auto Gordian Knot to create the audio files in it's AGK temp folder. The audio files show DELAY 2102ms when no such delay was requested.

Doom 9 shows a way to fix this in Virtual Dub which I looked up here:

http://forum.doom9.org/showthread.php?t=109724

But why am I getting this output in the first place? I see no setting in AGK to introduce such a delay that can be corrected. Is another setting in the program responsible for making this happen such as too small an avi output file request?

[nic2k4]

You didn't specify your source, so I'll assume it's DVD. This is likely caused by the way you ripped the DVD or any trimming you did before encoding in AGK.

I've had weird results like that converting '50 movies; try to use DVDHDdecrypter to rip the main movie only and don't trim anything. You can use Vdubmod set to direct stream copy to trim the AVI after.

I'm not sure why this happens, but I suspect somebody screwed up doing the original DVD encoding.

[loninappleton]

It may be a common problem in this way:

I noticed on two dvd jobs a similar effect. It happened on two separate encodes (from recent discs) where there is a separation between a Part One and a Part Two. The first had two episodes of a finale in one file. The second has an act break blackout indicating the intermission in a filmed stage play. The skew in audio happens on the second segment of both. This makes for some real head scratching until you can see the same effect happen twice.

VLC can be used to adjust sync on the fly with the J key (audio back) or K key (audio forward) in 50 millisend increments. (I just learned this recently.)

As to the original problem Virtual dub can split the file into two working segments and this is fine since there is a natural "found" break in these examples. Once the two files are made how can the Part 2 be remedied and saved? On one example I was able to sync the audio by hand to +1850 ms to bring the audio and image together using VLC keys as described above. Once that is known, can Virtual dub (or another tool) be used to fix the audio using the 1850 ms shift?

[Word; please avoid "basically" or "essentiallly". These words add nothing and have become tedious to see day in and day out.]

[manono]

It may be a common problem in this way:

I don't see how the problems are related at all. nic2k4 suggested (and I agree with him) that the problem in your first post is because of an improper decrypt. I would go further and guess that you didn't remove the unreferenced cells that led to the wrong delay being given. Your example in your second post doesn't have anything to do with an incorrect delay in the initial AC3 file.

Once that is known, can Virtual dub (or another tool) be used to fix the audio using the 1850 ms shift?

If you're still referring to the AVI made by AutoGK then, yes, you can direct stream the AVI AutoGK gave you and have a delay applied in the interleaving part of the audio tab.

[loninappleton]

Thanks for your considered and careful reply. We meet again.
The decryption job went in the usual way as described frequently by hello_hello and, from another thread, I follow that example. These are single files with two parts and not conjoined by me in anyway. The first half runs ok and then the second goes out of sync-- sort of like the space between was not accounted for. That's what makes it so odd.

As I recall, RipIt4Me does something with cell removal. Can you give more detail?

I can try RipIt4Me.

And though there is always a lot of discussion about the use of DVDFab Decrypter it has never worked for me. The current install I have completed gives a can't read the disk in drive D, an error which the DVD Fab site responds to by saying run Power DVD to load the disk and then run DVDFab Decrypter. This also did not work.

I will redo the disk with the RipIt4Me method and see if it has a cell structure removal
option. I also have the guide for that one.

Will report back if using RipIt4Me wizard mode solved the problem. It's worth going back and doing this over.

[manono]

Even in your first post you were only talking about a Part2 of a video? That's not what I understood from that post:

Just recently I've been having problems with audio synchronization using Auto Gordian Knot to create the audio files in it's AGK temp folder. The audio files show DELAY 2102ms when no such delay was requested.

So, the problem is only with the second part? Again, you can fix the delay after-the-fact in VDub. Or you can remove the delay entirely before feeding it to AutoGK using DelayCut. Sometimes those ways of doing it are easier than messing around with DVDFab HD Decrypter or FixVTS.

[loninappleton]

I was trying to be clear about the problem. Describing it in words is easily confused.
Simply put two distinct sections of an avi had different sound syncronization. The VOB set is continous and part of one AGK job session.

I've begun redoing the DVD with the RipIt4Me > DVD Decrypter and cleaning with FixVTS that the program does and encoding with AGK in an adequately large file size.

Delay Cut is a prog. I'm not familiar with. And with Virtual Dub I know only the demux procedures and such to do some audio filter procedures and then put the sound and video back together. Even that took a long thread to work out.

[loninappleton]

A quick update. Since the RipIt4Me and AGK procedure gave no difference and output avi is still out of sync, I'd need a plan to work on this through Virtual Dub or other tools.

What I know:

After the 'act break' on this live performace of a play from DVD, I can add 1850ms to the playback in VLC using the shortcut key. The keys J and K can adjust audio backwards and forwards and has a brief screen display so it can be done during playback without pausing video.

The best practice with plays for home viewing is to make two avi or other files separated at the act break. Many plays are done this way.

To procede then I could use Vdub to edit into two files and work on the sync of the second half only. Is this a good method to accomplish the goal or is there another?

I do not know how the programs discussed above do the synchronization/editing.

And tomorrow at some point I'm going to play the actual DVD chapters either on my stand alone player or in Power DVD to see if this is just a poor production job.

[manono]

I think I'm beginning to understand. Maybe. The whole thing is a single title (PGC) on the DVD and you encoded it as a single AVI. And the audio goes way out of synch for the second part.

Try opening the DVD in PGCDemux and ticking 'By Vob ID' rather than the default 'By PGC'. Then check the drop-down box on the right to see if the length of those two parts corresponds to two different Vob IDs. If so, you can also 'Check AV Delay' for them. You can also tick 'Single Cell' and see if there are any real short cells. If so, it's those that might be without audio and throw off the audio synch.

[loninappleton]

I think I'm beginning to understand. Maybe. The whole thing is a single title (PGC) on the DVD and you encoded it as a single AVI. And the audio goes way out of synch for the second part.

Yes. That is a good description. I viewed the second part last night.
The audio goes out of sync and stays consistent at the 1850ms delay described above. Once I manipulated VLC the audio stayed and played well with the new setting. I will do what you recommened and report back.

Try opening the DVD in PGCDemux and ticking 'By Vob ID' rather than the default 'By PGC'. Then check the drop-down box on the right to see if the length of those two parts corresponds to two different Vob IDs. If so, you can also 'Check AV Delay' for them. You can also tick 'Single Cell' and see if there are any real short cells. If so, it's those that might be without audio and throw off the audio synch.

[loninappleton]

Some things I have done:

I viewed the disk in the MPC player and the audio was in sync in the last act well within the second portion of the disk. I didn't duplicate the procedure on my stand alone player.

In PGC Demux I was confused about selecting menus or Title button.

What displays in PGCDemux from the DVD is Video TS IFO
VTS 01 0 .IFO
VTS 02 0 .IFO

Check AV delay gave all zeroes for multiple audio element (1-8 for instance)

I can post the VOB Id log file:

[General]
Total Number of PGCs in Titles=0
Total Number of PGCs in Menus=1
Total Number of VobIDs in Titles=0
Total Number of VobIDs in Menus=1
Total Number of Cells in Titles=0
Total Number of Cells in Menus=1
Demuxing Mode=by VOB Id
Demuxing Domain=Menus
Total Number of Frames=15
Selected VOBID=1
Number of Cells in Selected VOB=1
Selected PGC=None
Number of Cells in Selected PGC=None
[Demux]
Number of Video Packs=3
Number of Audio Packs=0
Number of Subs Packs=0
Number of Nav Packs=1
Number of Pad Packs=0
Number of Unkn Packs=0
[Audio Streams]
Audio_1=None
Audio_2=None
Audio_3=None
Audio_4=None
Audio_5=None
Audio_6=None
Audio_7=None
Audio_8=None
[Subs Streams]
Subs_01=None
Subs_02=None
Subs_03=None
Subs_04=None
Subs_05=None
Subs_06=None
Subs_07=None
Subs_08=None
Subs_09=None
Subs_10=None
Subs_11=None
Subs_12=None
Subs_13=None
Subs_14=None
Subs_15=None
Subs_16=None
Subs_17=None
Subs_18=None
Subs_19=None
Subs_20=None
Subs_21=None
Subs_22=None
Subs_23=None
Subs_24=None
Subs_25=None
Subs_26=None
Subs_27=None
Subs_28=None
Subs_29=None
Subs_30=None
Subs_31=None
Subs_32=None

ends

I saw nothing about comparing lengths in this Log File.

The Cell Times Logfile is empty-- no text.

It is likely I am not handling the program properly. In setup it defaults to menu but I used Title in
in the VOB Id Selection Domain.

This is all feeling around in the dark with PGC Demux-- I see no values to compare, so that's as far as I am.


For any progress give instruction in the form PGCDemux > selected IFO >Menu > plus the various options to get this right.

[manono]

What displays in PGCDemux from the DVD is Video TS IFO
VTS 01 0 .IFO
VTS 02 0 .IFO

Not the Video TS IFO, either 1 or 2, whichever the video is in.

In PGC Demux I was confused about selecting menus or Title button.

Title.

[loninappleton]

Today I am trying a workaround using a different program. I'll report on that as it progresses.

PGCDemux opens with a number of options preset. I have not messed with those.

In trying to gets logs from VOBid or Single cell, I'm only seeing one log. Is there a naming convention or some other thing needed to use the Process button and get logs for the things we are looking for? Also there was that empty Cell log mentioned above.

I do not know what things there are to look for in this prog. I don't see what to compare, or if found, how to make an edit and fix it to the 1850 millisecond difference to synchronize.

Playback seems to work alright. Something happens in the doing of it-- the encoding-- where a glitch must be picked up. It's also odd that this has happened to me twice (on different jobs) but just lately.

[hello_hello]

Maybe not solving the problem, but so you know how it works......

AutoGK effectively opens a set of vob files as a single file. DGIndex extracts the audio as a single stream (it simply appends the audio from from the next vob file to the previous one). DGIndex determines the amount of delay (relative to the video) used in the initial vob file. It writes the delay amount as part of the file name. AutoGK uses the specified delay when muxing the audio. You can't change that. You can only open the output AVI with VirtualDub and adjust the audio delay if need be, but 99.99999% of the time there's no need.

Did you create the entire set of vob files using a single title and DVDShrink? Or something similar?
It sounds like there's two vob sets involved which have been renamed in order to encode them as one. You didn't do anything like that did you? Because each vob set could use a different audio delay, but DGIndex doesn't know and only uses the initial delay, and therefore the second one isn't taken into account.

Have you tried opening the ripped files with DVDShrink and re-authoring it, or is that how they were created in the first place?
Are you re-encoding the audio, or using the original (it might indicate if there's non-audio data in the audio file which is being "skipped" when re-encoding it.

Anyway, if the problem's only occurred once, and the audio sync only goes out once in a particular spot you may be able to fix the encoded version rather than spending hours trying to work out "why".
Open the output AVI with VirtualDubMod and resva it as two individual AVI's, splitting it exactly where the audo goes out of sync. Open the newly saved second half, adjust the audio sync and resave it as a new AVI. Finally open the first half, append the "fixed version" of the second half, then save that as a new, single AVI.

[manono]

In trying to gets logs from VOBid or Single cell...

You don't need logs. Tick 'Title' and then either 'By VOB ID' or 'By Cell', open the IFO for the video, and use the drop down box to check for VOB IDs or cells, and maybe the 'Check A/V Delay' bar below to check for delays. Either that second part begins with a delay (and then it might have its own VOB ID with a delay) or the audio for the first part ends before the video does. That is, the audio for the first part might be shorter than its video. Either would account for your problem.

[loninappleton]

The drop down box business-- ok I'll try it now--- no I don't see any drop down boxes. A couple things allow you to browse.

I have the IFO loaded for VTS 01_0. How is the 'drop down box' opened? I see nothing like it. Is there a screen shot of what that looks like?

In the VOB id box 9 cells are showing but greyed out.

Now here is what makes me want to quit: time showing is only half the actual runtime and the other VOB-02 is just a brief screen less than a minute. Nothing is matching up in any of it.

I did manage to 'Process' the vob 01. I found the logfile but where is the rest?

These elements are coming from the dvd itself.

On the encode, no I have not done any audio treatments.

What I can do is go back to the method of using DVD Shrink and reauthoring the
main title in the way discussed on other occassions. Then use AGK to make the AVI.

The goal is to do what hello_hello says: to split the file into two parts as other similar
content for these plays has. Then do whatever is needed to the audio of the last part. No reassembly is necessary and no audio treatments either-- the audio is fine as is when in synch.

I not know why only half of the content is visible ( 1hr 19 mins 24 sec) for the disk when loaded in PGCDemux. This main title is 2 hrs and about 30 mins (not in front of it just now.)

My workaround did not seem to give any improvement.

Back to DVD Shrink and I report if another run through has improved anything.

[hello_hello]

If the ultimate goal is to keep the encoded version in two parts you can probably re-author it with DVDShrink that way. Even if the whole thing is a single title, use re-author mode and drag the title from the right pane to the left, then use the edit function to set the start and end points (1st half). Drag the same title from the right pane to the left a second time, then set different start and end points for it (2nd half). Use the backup function to resave the files and you should have two sets of vob files you can open with AutoGK for encoding. Each might use a different audio delay, but as it'll be encoded as two different encoding jobs it should be correct each time.

I don't know what's happening with PGCDemux as I've never used it.

[manono]

How is the 'drop down box' opened? I see nothing like it. Is there a screen shot of what that looks like?

It ain't rocket science.

[loninappleton]

It brings up another question though. I see that only one option is checked whereas mine opens with a bunch of those checked as default.

Since this PGCDemux program looks to be for those who have a lot of background in it, I may go the route suggested just above and work in DVDShrink.

The latest decrypt was done in DVDShrink with AnyDVD going in the background. A new avi was made in Auto Gordian Knot. I'm carefully watching this to see if the glitch has been reproduced again or not. But I am not up to that midpoint yet where the synch has gone out before.

[loninappleton]

Today I was trying to see any progress.

1. The new avi has the same problem after that blackout halfway point in the play.

There's what appears to be a good no sound place to make the cut and I copied down all the frame and time information on that.


AVI syth--- Jeez is there a simple diagram of button functions? About the only thing
I'm able to do is step through by frames with the mouse at the progress bar. And yes I've seen the AVI Synth wiki: another haystack trying to find a needle in it.

When opening up the VOB closest to the break I cannot detect any synch problem in playback. Summary: whole avi made from VOB's 1-6 has the synch problem-- an 1850 ms shift-- but the VOB 4 which has the break part way through does not.

Pardon me if I seem confused.

--> A long time ago I saw a diagram of the button functions of AVI Synth but my search for it has failed. Much of the wiki (scanning through the mountain of stuff in it) has to do with making scripts.

Perhaps making the cut and simply starting the second part will eliminate the synch problem. I'm beginning to wonder if this is a player problem: does skipping through to the desired point cause the audio to go out of synch?

In using players is it better to pause playbeack when using a slider to a different point?

[hello_hello]

It sounds like there's a gap in the audio at the point where the break is in vob4 (why, I don't know) which isn't being taken into account when encoding all 6 vob files (where the audio is all joined together). Either that or there's a problem with how the DVD was ripped.

Did you try re-authoring the DVD with DVDShrink as I suggested earlier? Use it's re-author and edit function to create a single title/set of vob files which go from the beginning to the "break point" in vob4, then create another set which go from what's currently the break point in vob 4 to the end. That'll give you two sets of vob files to encode individually.

It won't be a seeking problem when playing the AVI. For some reason there's an audio/video offset in the original files (break point in vob4) which isn't being taken into account when encoding. When you play the original vob files, the correct audio/video delay is used all the way through.

[nic2k4]

What Manono is looking for with PGCDemux is the presence of empty cells with a display time, during the intermission. If it's the case, say there are 6 cells with a 300 ms duration each and they don't display during playback; PowerDVD would play the movie just fine, but conversion software would account for those 1800 ms and cause the delay.

When opening up the VOB closest to the break I cannot detect any synch problem in playback. Summary: whole avi made from VOB's 1-6 has the synch problem-- an 1850 ms shift-- but the VOB 4 which has the break part way through does not.

If I get you correctly, you converted VOB 4 and didn't get the delay. In that case copy VOB 4-6 only to a separate folder and select VOB 4 as the input source in AutoGK. Caveat, if there are multiple audio tracks, you will only get the first one and if the video is 16x9 it may not display that way. Trim the beginning of the AVI in Vdubmod (set video to direct stream copy) to remove the end of the 1st part of the play and the intermission. Do the same to the end of the other AVI to save only the 1st part of the play.

[manono]

If I get you correctly, you converted VOB 4 and didn't get the delay.

I'm reading it slightly differently. I think he's saying that when playing the DVD or VOB 4 there's no delay. And, of course, there wouldn't be.

What Manono is looking for with PGCDemux is the presence of empty cells with a display time

Right, that's one reason I suggested earlier he check out the cells using PGCDemux. Very short audioless cells between parts 1 and 2 of the video (maybe in some black frames between the two parts) could easily cause the problem. It's also possible there are at least 2 VOB IDs in there. If the one for the second part has the delay then he'll at least know what causes it. But then, the use of PGCDemux for checking such simple things seems to be beyond him. If he comes up empty, then another other possible source of the problem is for the first VOB ID to have less audio than video (or, again, for there to be audioless cells in between). All these things are things to check when diagnosing a problem such as this one. Once the cause is known, the fix can be easily figured out.

And hello_hello's solution would also work. Cut the first part of the DVD and reauthor. Cut the second part and reauthor (leaving out the black frames (if any) between them), and then make the AVI. Or, as he also suggested, split the AVI in two and fix the delay that way, before rejoining the two parts. There are lots of ways to overcome the problem.

[loninappleton]

Some progress.

I used DVD Shrink in it's editing mode and (I think) made a Part One and Part Two.

Editing mode I finally found which opens the dialog for beginning and ending frames. The small icon button on the left panel I located by hovering over the icons above the DVD info box screen left.

Will see if that has straightened the problem.

--> As to PGC Demux, what is the method of finding the empty cells? Anything is beyond me that is not in step by step form-- all the steps. Case in point, having to play with DVD Shrink to get the editor to open. After all this time I had never seen it or know DVD Shrink had the capability.

[nic2k4]

And hello_hello's solution would also work. Cut the first part of the DVD and reauthor. Cut the second part and reauthor (leaving out the black frames (if any) between them), and then make the AVI.

I've tried that before and found it's hit or miss, it depends on what got messed up when the DVD was authored. I figured your path was better, didn't want to get this going in all sorts of directions.

There are lots of ways to overcome the problem.

That ever true.

[manono]

--> As to PGC Demux, what is the method of finding the empty cells?

Do you see the pic I posted before? Can you get to there by using that pic? That was an example using a DVD where the cells corresponded to chapters, so they're reasonably long. What you'll be looking for is cells that are much shorter. You said your delay is about 1850ms or 1.85 seconds. So, you're looking for a cell of about that length, or multiple cells (more likely) that might total that length or thereabouts. And while you're in there, also look for VOB IDs, with or without very small VOB IDs in between the two main ones.

[loninappleton]

I will follow up on what is said above now that I have a clue about it.

A quick peek at the slpit file done in DVDShrink looks like it will be free of audio synch problems. I can finally watch the Duchess get strangled in peace.

[loninappleton]

Today (but without going through the whole second part it looks like the method from hello_hello did the job and also gave a good starting point (music lead in intro plays from the beginning rather than a fraction of it.)

I will continue to look at that for errors and see the Duchess a few more times.

Working with PGCDemux I have, if I did it correctly, located a suspicious cell. It appears two places: in the Single Celll list and when checking AV Delay with the button lower right screen. The suspicious cell appears last which seems odd (as opposed to first) and reads 00:00:00:15.

The AV Delay gives a reading this time in the first element as Audio_1: 0x80 -> 2102 ms

which is at least a close number to my manual adjustment.

What to do next? Is there an editing procedure to follow (with steps) or is the information useless?
I agree that using this method looks to be more precise but the DVDShrink method once found seems a lot less obscure.

[hello_hello]

I've not worked with "problem" DVD files all that much so I can't give "fixing them" advice. The only reason I suggested re-authoring the vob files with DVD Shrink was because you said the ultimate goal was to keep the encode in two parts, so splitting the vobs into two sets before encoding made sense, and would maybe fix the problem. Normally, I guess you'd have to fix it "properly".

[loninappleton]

I have not responded to this and apologize.

The method of splitting that dvd file into two with DVD Shrink produced the results I used.

If there was an unused cell I can only say it must have been edited out.
I heard nothing further here about using the PGCDemux to... what...? hand edit the cell which method never was described.

I also continue to fiddle with Virtual Dum Mod to use that simple splitting technique.

For this problem it can be marked solved and may be useful to others.