PAL audio

[jchristl]

I've been attempting to convert a PAL AVI to NTSC DVD for some time now. I'm finding out that the audio is more trouble to convert than the video was.

The original AVI has a unique framerate of 25.0001 (per dvd2svcd) which I doubt matters. I ran the AVI through DVD2SVCD, and stopped it before CCE started, changing the Resize parameters based on what FitCD told me.

No problem there. I noticed the log file said "Undoing Nandub Blockalign" when it was extracting.

I've went and extracted the audio using various other techniques (dvd2svcd, avi2dvd, mplayer, etc). When I run the audio through besweet, besweet gui, or BeLight, the pitch seems to be always be higher.

I'm basically doing a 250 ms delay, PAL->NTSC 25000 29970.
The movie is 2:08:41
When I run this, the pitch is higher, it plays faster, and ends up 1:47:17 (or something).

I'm at my wits end.

BeSweet v1.5b29 by DSPguru.
--------------------------
Using hip.dll v1.19 by Myers Carpenter <myers@users.sf.net>
Using Shibatch.dll v0.24 by Naoki Shibata & DSPguru (shibatch.sourceforge.net).
Using SoundTouch.dll v1.21 by Olli Parviainen (www.iki.fi/oparviai/soundtouch).
Using MP2enc.dll v1.129 (23/5/2002), Engine 1.129 <http://www.cdex.n3.net>.

Logging start : 07/25/05 , 16:55:08.

BeSweet.exe -core( -input C:\work1\Audio\Extracted_audio_1.mpa -output C:\work1\Audio\Extracted_audio_1.mp2 -logfile C:\work1\Audio\Extracted_audio_1.log ) -azid( -s stereo -L -3db --maximize ) -ota( -d 250 -hybridgain ) -ssrc( --rate 48000 ) -toolame( -m j -b 256 ) -soundtouch( -r 25000 29970 )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : C:\work1\Audio\Extracted_audio_1.mpa
[00:00:00:000] | Output: C:\work1\Audio\Extracted_audio_1.mp2
[00:00:00:000] | Floating-Point Process: Yes
[00:00:00:250] | Source Sample-Rate: 48.0KHz
[00:00:00:250] +----- SoundTouch ----
[00:00:00:250] | Pitch : 0 semitones
[00:00:00:250] | Tempo : 20 percents
[00:00:00:250] | Rate : 0 percents
[00:00:00:250] | AA-Filter : Yes
[00:00:00:250] | Quick Mode : No
[00:00:00:250] +------- MP2ENC ------
[00:00:00:250] | Bitrate method : CBR
[00:00:00:250] | MP2 bitrate : 256
[00:00:00:250] | Channels Mode : Joint Stereo
[00:00:00:250] | Error Protection: No
[00:00:00:250] +---------------------
[02:08:41:746] Conversion Completed !
[02:08:41:746] Actual Avg. Bitrate : 213kbps
[00:09:48:000] <-- Transcoding Duration

Logging ends : 07/25/05 , 17:04:56.

I'm gonna try one more thing, before trashing this damn computer. Open the original AVI in VDub, say no to rewriting the header (it does ask), Direct Stream copy, save as AVI. Then See what I get.

:scared:

[Guest]

You can try DGPulldown for this conversion. The audio does not have to be touched at all.

[Prodater64]

DGPulldown is the best way to convert 25 to 29.97 fps.
http://neuron2.net/dgpulldown/dgpulldown.html


@neuron2: What about 29.97 true interlaced to 25?

[jchristl]

You can try DGPulldown for this conversion. The audio does not have to be touched at all.

Are you saying that all I gotta do is DGPulldown the video, and leave the audio alone? I'm gonna need to delay it about 250 ms. If that's true, that's great!!! :cool:

[Guest]

Are you saying that all I gotta do is DGPulldown the video, and leave the audio alone? I'm gonna need to delay it about 250 ms. If that's true, that's great!!! :cool: Assuming the source PAL is progressive, yes, that is true. If it is not, then you either have to decomb it (I say "decomb" rather then "deinterlace" to allow for one-field phase shift video that can be corrected with simple field matching) or use a conversion based on field blending. You'll have to allow for the audio delay in your encoding process, of course.

[jchristl]

Assuming the source PAL is progressive, yes, that is true. If it is not, then you either have to decomb it (I say "decomb" rather then "deinterlace" to allow for one-field phase shift video that can be corrected with simple field matching) or use a conversion based on field blending. You'll have to allow for the audio delay in your encoding process, of course.

@neuron2 - you're gonna get sick of me after all of this... :D

when I created a avs script with your telecine filter, it was telling me that it was progressive and interlaced. To me it seemed to be interlaced more in the lighter scenes and progressive in the darker scenes.

At any rate, When I simply loaded this avi into Vdub, and played one scene that has good lip movement. The scene starts out okay, but if I let it go for a few minutes the lips are off. This is without any processing - straight outta Vdub!!

I simply cannot understand why that is happening. At first I thought it was my video encoding, then I thought it was my audio encoding...
... now I don't know what to think.

:confused:

[Matthew]

Assuming DGPulldown away for a moment, the usual method for PAL->NTSC is to convert from 25->23.976 and then use pulldown on the video to get to 29.97. That leads to acceptable results if your avi ain't funky.

[Guest]

when I created a avs script with your telecine filter, it was telling me that it was progressive and interlaced. To me it seemed to be interlaced more in the lighter scenes and progressive in the darker scenes. There's no substitute for doing SeparateFields and then inspecting the fields by eye to determine whether material is progressive or not. Well, maybe having scharfis_brain do the inspection would be even more reliable, as I know from personal experience after making one hasty analysis right after waking up in the morning with a hangover!

At any rate, When I simply loaded this avi into Vdub, Sorry, but I have no idea what you mean by "this AVI".

and played one scene that has good lip movement. The scene starts out okay, but if I let it go for a few minutes the lips are off. This is without any processing - straight outta Vdub!! Please provide the full details about your source material and what process you applied. I'm just not as good at mind reading as I am at video processing. :)

[Guest]

Assuming DGPulldown away for a moment, the usual method for PAL->NTSC is to convert from 25->23.976 and then use pulldown on the video to get to 29.97. That leads to acceptable results if your avi ain't funky. Why assume something away that does the job faster and better? Your traditional method also requires progressive content.

[Matthew]

I assumed DGPulldown away because I was pointing out the fundamental flaw in his original approach - for informational purposes (I probably should have stated as such explicity).

I know the 'traditional' approach requires progressive content, my post was never meant to be a case of "for and against", or a suggestion that he not use DGPulldown. Rather, it was merely an explanation as to why things went so wrong in the first place.

As an aside, if dealing with a PAL avi of a movie that has non-pitch corrected mp3 audio, the time stretch option may be better, as a re-encode is required in any event.

[Paulcat]

First question: When your play the avi on your computer, is the audio and video in sync until the end? If so, you do not need to do anything to your audio, do a simple demux of the avi and save the audio as a separate file.

Process the video seperately if you can, leave it at 25 fps (progressive) and use DGPulldown to alter it after to 29.97 fps. If the running time of the audio and video are the same, all you need to do is author both files as a new dvd.

You didn't mention what type of audio was in the file. If it's mp3 you may have to use BeLight/BeSweet to convert it to mp2 or ac3 before you build your dvd

[jchristl]

Sorry, but I have no idea what you mean by "this AVI".

Please provide the full details about your source material and what process you applied. I'm just not as good at mind reading as I am at video processing. :)

:) Sorry, the source AVI that I am starting from. It is a PAL Divx capture. It is a 25.001 (oddly enough).

From GSpot:
AVI File Details
========================================
Name.........:
Filesize.....: 690 MB (or 707,182 KB or 724,154,368 bytes)
Runtime......: 02:08:41 (193,041 fr)
Video Codec..: DivX 3 Low-Motion
Video Bitrate: 601 kb/s
Audio Codec..: 0x0055(MP3) ID'd as MPEG-1 Layer 3
Audio Bitrate: 141 kb/s (70/ch, stereo) VBR
Frame Size...: 576x240 (2.40:1) [=12:5]

When loading it into DVD2SVCD it starts by saying
Undoing Nandubs 1152 BlockAlign value to 1
Redoing Nandubs 1152 BlockAlign value

When loading it into VDub, it says:
http://i14.photobucket.com/albums/a344/z-vap/Capture.jpg

At any rate, if I take this AVI untouched! and open it in vdub (no matter what I say at that previous message) if I go to any scene and watch the scene for a few minutes, slowly the audio goes out of sync with the video. If I stop it, and start up a totally different scene later in the movie, the audio will start insync, but go out of sync in a minute or two.

If I play it in media player classic, it's fine.

At this point, I haven't done anything to the file, yet.

[Guest]

Why don't you do what the message box says, i.e., decompress the audio to WAV and then recompress at CBR as MPEG audio or AC3?

[jchristl]

Why don't you do what the message box says, i.e., decompress the audio to WAV and then recompress at CBR as MPEG audio or AC3?

You mean Uncompressed PCM, then reconvert with Belight/Besweet to mpg? I could have sworn I tried that, but I can do it again, I guess.

I'll let you know...

[jchristl]

BeSweet v1.5b29 by DSPguru.
--------------------------
Using lame_enc.dll v1.32 (8/8/2003), Engine 3.90 <http://www.mp3dev.org/>.

Logging start : 07/26/05 , 16:04:46.

C:\Besweet\BeSweet.exe -core( -input C:\New Folder\movie.wav -output C:\New Folder\movie.mp3 -logfile C:\New Folder\BeLight.log ) -azid( -s stereo -L -3db --maximize ) -ota( -hybridgain ) -lame( -b 144 )

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : C:\New Folder\movie.wav
[00:00:00:000] | Output: C:\New Folder\movie.mp3
[00:00:00:000] | Floating-Point Process: Yes
Error 47: Error with LAME dll!
Quiting...
[00:00:00:000] Conversion Completed !

Logging ends : 07/26/05 , 16:04:46.

[Guest]

Try HeadAC3he.

[jchristl]

Try HeadAC3he.
I see nothing in HeadAC3he for bitrate, but I'm giving it a whirl...

(Urk... there it is, nevermind)

[jchristl]

Okey dokey... we're off to the races. Barring any audio delay adjustments, I'll let you know how things turn out. Authoring is in progress...

[jchristl]

Well, CBR encoding must have been the answer. It's seems to be a pretty good encode. There's a scene here and there that the lips/audio seem to be off by about 125 ms, but there are alot of scenes where the audio is right on.

I encoded it with HeadAC3he and 160 CBR. Thanks, neuron2 for leading my by the hand. :cool: :)