Error joining AC3 audio in .ts files

[Xoanon]

I downloaded a .ts broadcast that has, annoyingly, been split into two files (for reasons unknown). Both clips have AC3 audio. However, the first clip shows the audio as being 2-channel AC3 @192kbps. The second part says it has AC3 5.1 @384kbps. I figured I would try and join the two clips together, so I chucked them into Freemake, which showed both clips as having 5.1 AC3 @384kbps. Hmmm.

I then opened the two clips in tsMuxeR GUI and muxed them together. The AC3 is now precisely double the length of the program (the first half of the file is the audio, the second half is silence).

I tried demuxing the audio files from the two clips and focus on getting the video together in one file, which worked, figuring I would combine the two AC3 files later in eac3to. This also worked but again resulted in a file double the length of the video. The demuxed audio on clip 1 (the one that shows as 2.0 @192 or 5.1 @384 depending on the program) ends up being double the length once extracted on it's own.

Having demuxed the AC3 audio from the two original split files, I imported them into DVD Lab Pro to see how it would read them. Again, the first file showed as 192kbps stereo audio at double the length of the video it came from. The second AC3 file had the correct length and 5.1 channel configuation.

Why do different programs give me different channel configurations? The broadcast was recorded from HD with 5.1 audio but the split has caused the first file to show as stereo AC3 (in some programs) whilst the second file has its audio showing (correctly) as 5.1. Is there some kind of flag or stream error in the first clip that is causing the channel configuration to be misread? If so, why does Freemaker show it has 5.1 whilst everything else seems to show it as stereo?

I also tried opening/appending the two clips in AVIDemux. This resulted in a muxed file that was over 4x as long as the original program length.

I basically want to be able to combine these two .ts streams into one file with the original 5.1 audio but having tried a lot of options, nothing seems to be working. Does anyone have any ideas?

[ndjamena]

http://forum.doom9.org/attachment.php?attachmentid=11638&d=1285975545

Drag and drop the files onto that and it should tell what's what.

[Xoanon]

FileSize : 14608128 bytes
---------- First valid Header
Time eq. : 608672 ms.
SamplCod : 0 (0:48, 1:44.1, 2:32 KHz.)
BitRate  : 192 Kb/s
ChanMode : 2 (1:1/0, 2:2/0, 3:3/0, 4:2/1, 5:3/1, 6:2/2, 7:3/2)
FrameSize: 768 bytes
---------- Process ( 25.000000 fps is used for Trim)
Time: 0 ms.	Written: 23 frames 2.0	( 736 ms.)
Time: 736 ms.	Written: 9499 frames 5.1	( 303968 ms.)	Trim(18, 7616)
---------- End of File
Total time: 304704 ms. at EOF
T. written: 9499 frames 5.1.
T. written: 23 frames 2.0.

Thank you for that; I've never had to use this tool before - does that mean the first 23 frames of the video are stereo before it switches to 5.1?

[ndjamena]

It means there are 23 stereo frames in the stream. If the 5.1 file the program made seems complete you should use that from now on, many programs have issues with speaker layout changes mid stream.

[Xoanon]

I'm planning on converting the muxed .ts file into MPEG-2 for authoring. I've managed to mux clip 1 and clip 2 together in tsMuxeR (no audio). Now I have a single .ts file and two AC3 files; a fixed version of clip 1's audio (with the first 23 frames removed, leaving me with the correct 5.1 audio) and clips 2's audio. I can combine the two together in eac3to but the audio will be 736ms out. I assume I can correct that in DVD Lab Pro using the Audio Delay option?

[ndjamena]

I have no idea, all you really need is 23 silent 5.1 ac3 frames to fill the gap, which shouldn't be hard to manage one way or another.

[Xoanon]

How do I insert those silent frames, do you know? I have never encountered this problem before, that's all. I thought I'd seen it all until today.

[poisondeathray]

You would create a silent 5.1 audio clip 736ms long, then append that to the 1st audio (which you already cut the stereo section off with something like delaycut)

[Xoanon]

I didn't need to use delaycut to edit the dodgy AC3 file, as SplitAC3 seems to have created two new AC3 files from the part 1 clip; one thats 736ms long (and stereo) and a second one being the rest of clips 1's audio (in 5.1).

Could I create a new silent 736ms long 5.1 AC3 file, and append that to the second part of clip 1's audio, and then append that to clip 2's audio (which has always been correct) in eac3to? Ergo I would have one AC3 file of the entire .ts program audio?

[poisondeathray]

Yes - Isn't that what you want? A single audio file to match the appended video ?

[Xoanon]

Oh Yes, I just wondered if I could do it in eac3to specifically instead of delaycut.

eac3to does not recommend keeping the dialogue normalization settings, but if I don't keep them, the audio is boosted too loud. Is there any real reason to turn it off?

[poisondeathray]

If it's already cut by that other program then there is no need to use delaycut

dialnorm is more of a subjective discussion, really it's up to you

[Xoanon]

Thanks for all your help, guys.

To recap, I muxed the two video clips, clip1 and clip2 into one single (and silent) .ts file, which I need to convert to MPEG-2 for DVD burning.

I had the faulty clip1 stereo/5.1 AC3 audio, and the clip2 5.1 AC3 audio. Clip1 was dragged onto splitac3, which created a 736ms stereo AC3 file (which I'll call clip1a, which I discarded) and the rest of clip 1's audio in 5.1 (clip1b). I've created a silent 736ms 5.1 audio to replace the original clip1a, onto the end of which I added clip1b and clip2. This, in theory, should be the full programme audio and will sync 1:1 to the .ts file. I won't be able to check until later, but if I run into any issues I will post back here.

As it stands, the appended AC3 file is 45:45 exactly, whilst the .ts video file shows as 46:46:240 in VirtualDub. Not sure why there is some discrepancy, there.

[ndjamena]

Well, I noticed you're American, yet left the program set to PAL, but if that was the problem the damage would be more severe...

...or non existent, depending on what this means:

Where optional fps is used to calculate video Trim() for 5.1
Valid values between 1.0 and 1000.0
By default 25.0 fps is used.

How are you loading it into VirtualDub? I don't think .ts files have an actual index...

[Xoanon]

The source material was PAL so I opted to leave it that way. As for the Vdub thing, I have an MPEG-2 filter installed so I assumed that's how it opened?

After doing everything above, the audio was out of sync (due to it being shorter than the video, for some reason) so I gave up. However, it might be useful to know why this might have happened for future reference. I've had no issue with single-file .ts videos, but this split caused a right headache