NTSC to PAL DVD Audio

[mattstan]

Hi,

While I was away over the weekend (it needed 20+ hours on my PC) I used Procoder to convert a NTSC DVD to PAL. I had used VobEdit to demultiplex the DVD first to get a M2V video file and AC3 audio file to convert with Procoder. The M2V video came out as PAL no problem but the WAV PAL sound file contained no sound at all.

So now I want to convert the NTSC AC3 audio file to PAL so I can remultiplex the files and burn to DVD. I've been trying to do this with BeSweet but can't seem to get the conversion settings right. I'm pretty sure one attempt came out as PAL but DVD Author said it was not ok with this error:

"For a standard DVD you can only use the following formats
Dolby Digital (AC-3), MPEG-1 Audio Layer-2, or Linear PCM"

The file had an AC3 extension but I guess I must have gone wrong somewhere.

Any advise would be great on how to do this with BeSweet or some other software (preferably free like BeSweet), I don't think Procoder will allow me to convert just the audio (not that I can work out anyway) and to restart the whole process would take too much PC time.

Thanks.

[violao]

If you have stereo then you can use besweet with:

besweet -core( -input input.ac3 -output output.wav -2ch ) -ota( -r 23976 25000 )

At this point I would also use -azid ( -g max ). Now you can encode with your favorite ac3 encoder. You can do this with besweet or headAC3che, but I don't think ac3enc volume issue is solved.

If the sound is mono then first convert to wav with lame:

lame --decode ...

then resample with soundstretch:

soundstretch input.wav output.wav -t=4,2709376042709376042709376042709

This number is (25000/23976 - 1)*100. BTW, I believe more precise results are obtained by using 1001/960 ratio.

Have fun!

[Paulcat]

Did your conversion change the running time of the M2V file? If the original was 120 minutes and your new one is STILL 120 minutes (not 4% shorter), then you can remux your original AC3 file.

[mattstan]

Originally posted by violao
If you have stereo then you can use besweet with:
besweet -core( -input input.ac3 -output output.wav -2ch ) -ota( -r 23976 25000 )
Have fun!

Many thanks I'll have a go later.

[mattstan]

Originally posted by Paulcat
Did your conversion change the running time of the M2V file? If the original was 120 minutes and your new one is STILL 120 minutes (not 4% shorter), then you can remux your original AC3 file.

Thanks for the info.

The M2V file is 4% shorter (plus one second, unless I made a maths error). How can this be ok? The film is about 100 mins, so my PAL version is 4 mins shorter -- that is a 'different film' in my book! Can you convert to PAL without this discrepancy and if so how?

Cheers.

[Paulcat]

Originally posted by mattstan
Thanks for the info.

The M2V file is 4% shorter (plus one second, unless I made a maths error). How can this be ok? The film is about 100 mins, so my PAL version is 4 mins shorter -- that is a 'different film' in my book! Can you convert to PAL without this discrepancy and if so how?

Cheers.
The difference between an NTSC film and a PAL one is that the PAL format is larger and it is sped up slightly to get 25 fps. Which means you will have to time-stretch (or shrink in your case) the audio by 4% to keep it in sync.
If you are converting to PAL, you could try converting the video portion to PAL size and keep the NTSC framerate, then use the program "DGPulldown" (search the forums here) to change the flags to 25 fps, and then remux your audio and video.
I don't know if it will work, all my conversion has gone the other way!

[violao]

Originally posted by mattstan
The M2V file is 4% shorter (plus one second, unless I made a maths error). How can this be ok?

It isn't exactly 4%. It's 4,09590.... (960/1001 I believe, please correct me if I'm wrong). 24 -> 25 is 4%. Also audio might be shorter, so it isn't reliable just to compare durations. You should mux and preview beginning and end chapters.

[mattstan]

Originally posted by violao
If you have stereo then you can use besweet with:
besweet -core( -input input.ac3 -output output.wav -2ch ) -ota( -r 23976 25000 )
Have fun!

Hi,

I tried this last night with this command line:

BeSweet.exe -core( -input "j:\DVD\in.ac3" -output "j:\DVD\New.wav" -2ch ) -ota( -r 29970 25000 ) -azid ( -g max )

Note that my NTSC is 29.970 not 23.976 and so I changed my rate like this:

Yours: -r 23976 25000
Mine: -r 29970 25000

The resulting audio file was still approx. 103 mins (the same as the NTSC file) and not approx. 99 mins (the length of the PAL video).

In case it may help here is BeSweet's output:

[00:00:00:000] +------- BeSweet -----
[00:00:00:000] | Input : j:\DVD\in.ac3
[00:00:00:000] | Output: j:\DVD\New.wav
[00:00:00:000] | Floating-Point Process: No
[00:00:00:000] +-------- AZID -------
[00:00:00:000] | Input Channels Mode: 2/0, Bitrate: 192kbps
[00:00:00:000] | Output Stereo mode: Dolby surround compatible
[00:00:00:000] | Total Gain: 3.423dB, Compression: None
[00:00:00:000] | LFE levels: To LR -INF, To LFE 0.0dB
[00:00:00:000] | Center mix level: BSI
[00:00:00:000] | Surround mix level: BSI
[00:00:00:000] | Dialog normalization: No
[00:00:00:000] | Rear channels filtering: No
[00:00:00:000] | Source Sample-Rate: 48.0KHz
[00:00:00:000] +-------- FRC --------
[00:00:00:000] | Source Frame-Rate: 29970
[00:00:00:000] | Dest. Frame-Rate: 25000
[00:00:00:000] +---------------------
[01:43:52:224] Conversion Completed !
[01:43:52:224] Actual Avg. Bitrate : 463kbps
[00:45:29:000] <-- Transcoding Duration

Many thanks.

[violao]

Can't comment on this, but please note that both 29.97 and 23.976 NTSC video duration is the same. There is no speedup or slowdown between the two. 29.97 is being produced from 23.976 by telecine process that involves duplicating fields without any speedup. As a consequence audio saved from both 23.976 or 29.97 would last for the same time. If you want to make PAL from this you need to speed it up from 23976 -> 25000 regardless of the frame rate. Try soundstretch. I have very positive experience with it.

[zilog jones]

Umm, changing the audio speed from 29.97 fps to 25 is only needed if you're changing the framerate directly - which is a really lame NTSC-to-PAL conversion IMO.

If the source is video, the conversion from 59.94 to 50 fields should result in the same length... more or less. Audio conversion shouldn't be needed.

If the source was film, you need to do the 4% speed change as said before. Some programs may not recognise NTSC 3:2 pulldown video as 23.976fps and will say it's 29.97, but it's supposed to be 23.976.

What is your source? If it is interlaced video and your video is now a different length, you or the program have done something seriously wrong...

[mattstan]

Originally posted by zilog jones
If the source was film, you need to do the 4% speed change as said before. Some programs may not recognise NTSC 3:2 pulldown video as 23.976fps and will say it's 29.97, but it's supposed to be 23.976.

What is your source? If it is interlaced video and your video is now a different length, you or the program have done something seriously wrong...

Many thanks.

You're right Procoder said it was 29.97 but even a different info display of the same software said it is actually 23.97, which I have confirmed.

So I tried this BeSweet command:

BeSweet.exe -core(
-input "j:\DVD\in.ac3"
-output "j:\DVD\out.wav" -2ch
-logfilea "j:\DVD\bs.log" )
-azid( -g max )
-ota( -r 23976 25000 )

Unfortunately the resulting wav came out at the same lengtyh as the NTSC file at 103 mins instead of the same length as the PAL file of 99 mins (both min lengths approx.).

Any ideas what's going wrong?

Thanks guys.

[violao]

No, but at this point I would try step by step approach. Decode with:

azid [-g 3.423] in.ac3 out.wav
(sorry I mentioned lame in previous post, as if your sorce was mp3)

then:

soundstretch out.wav stretch.wav -t=4,2709376042709376042709376042709

or perhaps -t=4,2708333... (I'm not sure)

If my calculation (1001/960) was right this should end up as 1:39:36.958. Note that soundstretch doesn't change pitch, while besweet -ota(...) does.

[mattstan]

Originally posted by violao
No, but at this point I would try step by step approach. Decode with:

azid [-g 3.423] in.ac3 out.wav
soundstretch out.wav stretch.wav -t=4,2709376042709376042709376042709

or perhaps -t=4,2708333... (I'm not sure)

If my calculation (1001/960) was right this should end up as 1:39:36.958. Note that soundstretch doesn't change pitch, while besweet -ota(...) does.

Many thanks again for your help.

The step-by-step approach worked but not with soundstretch.

When I used azid [-g 3.423] with the with the -2ch -core switch to get the NTSC wav, and soundstretch exactly as you wrote IE.

soundstretch SSIN.wav SSOUT.wav -t=4,2709376042709376042709376042709

and finally converted soundstretch's output to AC3 using:

BeSweet.exe -core( -input "SSOUT.wav" -output "SSOUT.ac3" )
-ac3enc( -b 448 -6ch )

I ended up with a 'PAL' AC3 that was 15 seconds too long and consequently out of sync (it got more out of sync the further you got into the film).

However the process worked using just besweet but doing each stage in turn. Like this:

1) Convert original NTSC AC3 to NTSC WAV:

BeSweet -core( -input "NTSC.ac3" -output "NTSC.wav" -2ch )
-azid( -g 3.423 )

2) Convert frame rate from NTSC WAV to PAL WAV:

BeSweet -core( -input "NTSC.wav" -output "PAL.wav" -2ch )
-ota( -r 23976 25000 )

3) Convert PAL WAV to PAL AC3:
BeSweet -core( -input "PAL.wav" -output "PAL.ac3" )
-ac3enc( -b 448 -6ch )

Then I just re-authored the DVD using my already converted DVD video PAL M2V file and the newly converted to PAL AC3 file.

There is one question remaining that I have, because I don't understand what it did, and that is what was the purpose of the -azid( -g 3.423 ) switch in stage 1:

BeSweet -core( -input "NTSC.ac3" -output "NTSC.wav" -2ch )
-azid( -g 3.423 )

What did this do and why not just use 'max' instead? Like this:

BeSweet -core( -input "NTSC.ac3" -output "NTSC.wav" -2ch )
-azid( -g max )

Many thanks once again.

[violao]

Because you already measured max gain value in last attempt:

[00:00:00:000] +-------- AZID -------
[00:00:00:000] | Input Channels Mode: 2/0, Bitrate: 192kbps
[00:00:00:000] | Output Stereo mode: Dolby surround compatible
[00:00:00:000] | Total Gain: 3.423dB, Compression: None

So there was no need to do that again. With -g max you would have 2 azid passes which is longer and not necessary once you have this value.

Re besweet, this is probably something for the author. I remember I had similar problems using 1-pass besweet with -soundtouch, but never with -ota.

[mattstan]

Originally posted by violao
Because you already measured max gain value in last attempt:

So there was no need to do that again. With -g max you would have 2 azid passes which is longer and not necessary once you have this value.

Re besweet, this is probably something for the author. I remember I had similar problems using 1-pass besweet with -soundtouch, but never with -ota.

Thanks again violao.

Let me check I've got this right.

You'd seen in my earlier posted output this line:

"Total Gain: 3.423dB, Compression: None"

So you knew the 'max gain' (is that the correct terminology?) was 3.423Db, so this value could be used in the -azid switch instead of max because 3.423 is the max and using the value instead of max saved time because BeSweet would do only 1 pass instead of 2. Is that correct?

Many, many thanks for all your help, I really appreciate all the time you've spent helping me -- I wouldn't have got this going without your help.

Regards,

Matt

[violao]

Correct. This is particularily important when you use ac3enc for ac3 encoding since it is known to lower the volume somewhat and doesn't use DRC nor dialog normalization, so without normalization at the beginning you would have probably produced very low volume soundtrack.

No problem at all Matt. I'm glad it worked for you.

[mattstan]

Originally posted by violao
Correct. This is particularily important when you use ac3enc for ac3 encoding since it is known to lower the volume somewhat and doesn't use DRC nor dialog normalization, so without normalization at the beginning you would have probably produced very low volume soundtrack.

No problem at all Matt. I'm glad it worked for you.

Thanks again for the explaination, it's good to understand these things beyond just getting it to work. I've created a how-to file with all the processes and helpful info, so next time it'll be easy. And maybe I in turn can help someone out in the forums on the subject, after all that's what makes newsgroups and forums so great.

Thanks,

Matt