timecode problem with ac3 and Maestro

[shorton]

Hi guys:

I'm creating a technical audio track (beeps at precise moments syncronized with a pattern on the video) for a DVD I'm making. I create the audio as wav in Adobe Audition. I then encode it with Sonic Foundry Soft Encode to ac3.

I also create my video using avisynth and CCE. The video is created with a 29.97 frame rate FWIW.

I author using DVD Meastro. When I add my ac3 file to the Maestro timeline, I get a warning that says "the Timecode of the video does not match the timecode of the audio" (or something very close to that). I'm creating the audio file exactly as long as the video based on 29.97fps or .033367sec/frame, set number of progressive frames (i.e. 290 frame video = 9.676343 secs)

I can't figure out what I'm doing wrong or how to avoid the error/warning. I have control of the whole process but don't know what to change.

Can anyone tell me how to fix it?

Thanks,
Scott

[bigotti5]

In Maestro check
File - Preferences - Movies - Ignore AC3 Embedded Timecode

or in Sonic Foundry Soft Encode uncheck
Options - Encode Settings - Time code - Add time stamps

[shorton]

In Maestro check
File - Preferences - Movies - Ignore AC3 Embedded Timecode

or in Sonic Foundry Soft Encode uncheck
Options - Encode Settings - Time code - Add time stamps
Thanks. But, I do have "Add time stamps" checked and I still get the warning. It doesn't prevent me from compiling.

I'd like to know the timecodes are synced. I think. I'm trying to create a "lip-sync" test pattern so getting the audio *perfectly* in-sync with the video is my primary objective.

Telling Maestro to ignore it dosen't seem to fit that plan. I was under the (possibly incorrect) impression the timecodes would help ensure the audio and video would be in-sync. If I tell Maestro to ignore it, perhaps I could just be using a wav file instead?

Is there some way to get the timecode created in Soft Encode to match the timecode in the video file?

I have found that if I import my wav into Adobe Premiere and encode with the Minatoga (or whatever it's called) dolby encoder, I don't get the sync warning, if I use the matching video segment created in Premiere (which I don't want).

[bigotti5]

I do have "Add time stamps" checked

or in Sonic Foundry Soft Encode uncheck

Timecode in SonicFoundry is 30 fps (non-drop) and not 29.97 fps (drop) so they cannot match

Maestro will do a good job without sync problems

[shorton]

Oops, missed the "un". Sorry about that. Thanks for pointing it out.

Perhaps if I encode my video at 30fps that woudl help, but isn't DVD (i.e. Maestro) gonna do 29.97?

The 29.97 vs 30 and drop vs non-drop thing has had me scratching my head. Wish I knew where to read more abou it. I had tried 30fps in CCE but you have to turn off the target of DVD to do it.

I'll try turning the timecode off as instructed.

But, since I'm shooting for exact times, I'm concerned "good" will be off. I'm stuck with Maestro, I could use a different audio encoder (i.e. beesweet), or the one in Premiere Pro. Given that I'm using Maestro for authoring, what would you recommend for exact sync in my video and audio encoding?

Thanks,
Scott

[bigotti5]

Imho AC3 timecode has nothing to do with Audio/Video sync in Authoring. Maestro will mux in correct way and generate correct PTS (Presentation time stamps)

AC3 timecode is for synchronizing different machines in a hardware SMPTE environment

[shorton]

Imho AC3 timecode has nothing to do with Audio/Video sync in Authoring. Maestro will mux in correct way and generate correct PTS (Presentation time stamps)

AC3 timecode is for synchronizing different machines in a hardware SMPTE environmentGreat. That is encouraging. So I'l concentrate on getting the wave built to teh correct length.

Since I've been using 29.97fps, I am calculating my audio length as follows: 1/29.97 = 0.033367 sec/frame. So for example if my video was 10 seconds long, I would make the audio 3.3367 secs long to match exaclty (well very very close).

Right?

Thanks very much for the time to help.

[tebasuna51]

Since I've been using 29.97fps, I am calculating my audio length as follows: 1/29.97 = 0.033367 sec/frame. So for example if my video was 10 seconds long, I would make the audio 3.3367 secs long to match exaclty (well very very close).

Right?
I don't understand. If your video is 10 seconds long, you need a audio 10 seconds long. If you put only 3.33 sec. is normal the Warning.

To be more exact:

If your video have 300 frames, and is played at 29.97fps, the duration is 10.010 sec.

The audio ac3 stream have also fixed frames of 0.032 sec. (48 KHz) then, if you have 313 audio ac3 frames (48 Khz) the audio duration is 10.016 sec. This duration is independent of video framerate and is always 10.016 sec. long in a PAL system (25 fps) or a NTSC system.

[shorton]

Since I've been using 29.97fps, I am calculating my audio length as follows: 1/29.97 = 0.033367 sec/frame. So for example if my video was 10 seconds long, I would make the audio 3.3367 secs long to match exaclty (well very very close).DOH! Not enough sleep I guess, sorry about that. What I MEANT to say was:

Since I've been using 29.97fps, I am calculating my audio length as follows: 1/29.97 = 0.033367 sec/frame. So for example if my video was 100 frames long, I would make the audio 3.3367 secs long to match exactly (well very very close).

If your video have 300 frames, and is played at 29.97fps, the duration is 10.010 sec.

The audio ac3 stream have also fixed frames of 0.032 sec. (48 KHz) then, if you have 313 audio ac3 frames (48 Khz) the audio duration is 10.016 sec. This duration is independent of video framerate and is always 10.016 sec. long in a PAL system (25 fps) or a NTSC system.So this means one cannot get a "perfect" sync? In your example I'd be off by .006. Perhaps that's not a big deal. Is it the AC3 file that's the issue or the 48Khz? That is, can I use a wav (vs AC3) and get perfect sync?

My video sequence has 28 frames where there should be silence, and one frame where there shoudl be a beep. If my beeps are to be 1 AC3 frame in duration (.032 sec) which is close to one video frame. My beeps need to occur after about 28 video frames of silence so 0.934s. So using the .032 audio frame the closest I can get is 29 audio frames (0.928) .006s off.

On the second iteration of the video I'll be at 57 frames of video (cumulative from beginning) 1.902s. This is ~59 audio frames or 1.888s which is off .014s.

As my video iterations get longer, the audio start (based on it's frame rate) will dance around the exact sopt on the video where teh tone will start. No way to get (AC3) literally exact then?

[tebasuna51]

Not problem for the exact sync. You can make a total exact wav and after encode to ac3. Only if you make partial encodes to ac3 in order to append after you can get sync problems.

Only the total length is forced to be a exact number of frames and then the time length must be a 32 ms multiple. Inside a ac3 frame you can have 20 ms of silence and 12 ms (for instance) of beep.

The frame size for 48 KHz is 32 ms, for 32 KHz 48 ms and for 44.1 KHz 34.8 ms (not exact, there are two frame sizes to match exactly 44.1 KHz)