Audio Out of Sync—18 Hours of Encode Time Wasted

[Tolwyn]

Project plays fine in Sony Vegas.
Frameserved out to TMPGEnc 3.0 XPress using DebugMode Frameserver.

To use the FrameServer, you have to render the timeline to an AVI file using the
"DebugMode FrameServer" file type.

Screenshot 1
Screenshot 2

(dummy file used here just so you can see the settings...)

If I check the box "Write audio as PCM samples in signpost AVI" (or whatever), then the encoding application thinks the entire stream is almost 5 or 6 hours long. And using that option seems to only be for seeking, which I didn't deem as important.

My entire movie is out of synch.
The audio starts right away, when it should only start when the video stream starts. See Screenshot 3.

What would have caused this? The audio shouldn't have begun until the 100th frame or so, yet in my finished encoded Mpeg file, it starts at frame 1.
Obviously, there's no audio on the track where the project title starts. Nor is there any audio information when the project credits role.


???

In the Cut Edit section of TMPGEnc, it plays back correctly (of course, since it's being frameserved, it can't play/seek the whole way through, but the initial several hundred frames are correct).

How can I repair this and how can I avoid it happening again? The audio stream itself initially wasn't VBR, it was standard Type2 DV PCM (uncompressed as far as I know).

What did I do wrong?

ARRRGHHH!!!!!

[Tolwyn]

So I suppose I can load up the MPG in VirtualDubMOD, extract the WAV audio, add 4 seconds and 9 frames of silence to the file, and then multiplex it back out using TMPGEnc, right?

I'm not sure if I said that right, but is that possible at least? I'd really have to guess at the time offset. Good grief. I'm really discouraged right now. 18 hours is a helluva long time to endure only to have the final product porked.

[Tolwyn]

Arrgh! I can't even do that. It won't let me save WAV audio. !#$%!!

[reboot]

Demux in tmpgenc.
Adjust audio in Goldwave.
Author, burn.

[zeus163]

I don't know about Vegas Video, but in Premiere you can export just the audio. If you export the audio and mux that with the video file you have already created perhaps that will get you back in sync. (you'll have to demux the already encoded file of course).

I use the same frameserver and have had only one audio problem. When that happened, I exported the audio, created an AC3 file, and muxed it with the video. That worked.

I usually use the frameserver with CCE or Procoder, but I would think that shouldn't be any different than this.

Good luck. I feel your pain.

[Tolwyn]

Now I'm not sure what I need to do when Muxing.

How do I ensure that it's going to use 224kbps in the WAV file?

How do I enocde to MP2 from WAV?
I searched for a tool (BeSweet), installed it, and it refused to work right, or I am too ignorant to use it. Even tried the GUI and when I clicked "GO!!" it just popped up an empty notepad window.

Arrgh.

[trock]

Are you using TMPGEnc's Cut Edit section to remove the extra unneeded hours at the end?

I assume you are but I thought I'd mention it just in case. I use the same programs and system as you and haven't had any sync problems and I always remove the extra in TMPGEnc before encoding.

[reboot]

Tmpgenc will encode 224kbps audio from .wav
Depending on your authoring application, you may not need to remux.
Demux in tmpgenc.
Adjust your audio in Goldwave.
Save .wav
Encode audio in tmpgenc.
Author, burn.
Deja vu?

[Tolwyn]

There's really no "hours" at then end. It's almost a misreporting from the frameserver to the encoder.

Are you using TMPGEnc's Cut Edit section to remove the extra unneeded hours at the end?

I assume you are but I thought I'd mention it just in case. I use the same programs and system as you and haven't had any sync problems and I always remove the extra in TMPGEnc before encoding.

[Tolwyn]

Not deja vu. Just you have to speak a little more slowly to me. I'm kinda slow.

THIS reply you gave has a bit more granularity to it, which helps me a great deal.

I didn't know I could just encode a WAV file from within TMPGEnc.

Tmpgenc will encode 224kbps audio from .wav
Depending on your authoring application, you may not need to remux.
Demux in tmpgenc.
Adjust your audio in Goldwave.
Save .wav
Encode audio in tmpgenc.
Author, burn.
Deja vu?

[trock]

There's really no "hours" at then end. It's almost a misreporting from the frameserver to the encoder.

Yes, but if I don't cut them off, it adds hugely to the encoding time.

[Tolwyn]

trock: there's no way to do that.
reboot: My version of TMPGEnc 3.0 XPress will not handle WAVE files.

Only video files.
I'm back to square one.
Looks like another 18 hours.

[Tolwyn]

This is really confusing.
In my project...
The video/audio "stream" (clip) starts 4 seconds and 10 frames into the project.

Extracting the audio and inserting 4 seconds and 10 frames of silence looks right, but in encoding (multiplexing) it's still off.

What is going on?
Why would this happen?

[trock]

trock: there's no way to do that.

LOL, we must be operating in parallel universes. I do it practically every day and it reduces my encoding time by 4-5x.

[reboot]

Get tmpgenc. Not 3.x express. Even old version 12b (free) will do.

[Tolwyn]

trock. I'll have to take a screenshot. In cut edit mode, it's the normal time of the compilation (meaning it shows 1 hour, 27 minutes or so).

But when I'm on the encode page, and entering the VBR, it is reporting that the audio portion of the stream is 5 hours.

There is no way to "cut" this mis-reporting.

reboot... the free version expires, doesn't it?
I thought 3.x was everything 2.x was. Damn those Pegasys bastards.

[zeus163]

Yes, you can deumux in TMPGENC. That's the first step I'd take with the completed file. Then I'd export the audio file from Vegas Video at 48K. From there, I pull the file into BeSweet and use the GUI to set it at 224k and then do wav to ac3. Doom9 has a pretty good guide for BeSweet and if you go into the forums you can try out BeLight which a lot of people like since the GUI for BeSweet is fairly complex (but that's what the guide is for). Instead of taking another 18 hours encoding, read up on BeSweet or BeLight and who knows, maybe that will solve your problem.

I'm not sure what is going on and why it is out of sync. I feel your pain, but you may not have to do another encode if you can use those AC3 audio encoding programs.

Good luck.

[trock]

Ah, then we're talking about 2 different things. In mine, in the cut edit mode a 30 second clip shows as 2 hours and 5 mins and encodes accordingly. If I use cut edit to make it the correct 30 seconds, it encodes in the correct time (4 to 5 times less than if I don't cut the end off). But this is different than what you're running into.

[Tolwyn]

Zeus... how do I export the audio from Vegas Movie Studio? I've looked all around the application and can't do it.

I do, however, see what has happened and why... more in a minute...

[trock]

Render as a wav file in Vegas (mute the video).

[Tolwyn]

-edited-
Found it.
Thank you.

[Tolwyn]

By the way, I truly appreciate and value the time and energy you folks are expending to my issues here.

I'm still going to followup with the problem to perhaps warn others, or to establish if *I* did something the "wrong way..."

[zeus163]

I don't use Vegas, but I see that Trock has found the answer for you.

[Tolwyn]

I'll have to use MP2, not AC3, as the video bitrate was established to "fit" on the DVD at 224kbps. No worries there, though.

I updated BeSweet, and downloaded the updated GUI.

On an aside, is there an MP2 codec that I can download/install that will be recognized by Adobe Audition 1, 1.5 or Cool Edit 2.5?

[Tolwyn]

Yeah. That audio rendering thing is tucked away. It's obvious once found, but it would be nice if it was included on the tool bar or something.

I don't use Vegas, but I see that Trock has found the answer for you.

[Tolwyn]

Ok. I don't know if this is a Vegas issue, or a Frameserving issue... or, quite frankly, if it's a TMPGEnc issue.

I can't explain this very well, but I'll show you the pictures.

In project playback, the audio starts as it should. It begins at frame 00:00:04:10... even though the clip is "cropped" or "edited" to trim off some of the beginning stuff. Now, note that there is no wave audio explicitly added from frame 1 to frame 130. So, I don't know how an encoder treats this "nothing" until stuff actually starts...? Or if it's the frameserver's job to understand that there's nothing "there" and just generate silence... I am struggling in articulating this problem...


Now, obviously, there is video and audio "data" that comes before 00:00:04:10, it's just "supressed" (cropped?) away. (Can't think of terminology here)... but as you can see... the audio does indeed go all the way back to frame 1.

The encoder or frameserver can't figure this out, and begins encoding the audio at frame 1 in the MPG file, not only that, but also, encodes the audio that was CROPPED AWAY!!!