AC3 Sound Issues

vbimport

#1

Chickeman and others

I have been meaning to ask this for a long time but have never got around to it until now.

Every time I backup a movie or convert DIVX to DVD-R I always find that the sound plays a lot softer on the copy.

For example I when I backup a movie using Chickenman’s guide (which is the best on the net by the way) everything is perfect except when I play back the copy I need to turn the sound up a lot higher than when playing the original.

This has happened on all 180 of my movies.

Am I doing something wrong.

When DVD2SVCD finishes and you are left with an .ac3 file, I then use AC3 machine to make it compliant and then add it into DVD Author.

Any ideas?


#2

AC3Machine is a GUI front-end that uses BeSweet for audio conversions. The author of BeSweet gives an explanation of the low volume here.

Regards
TZ


#3

That link to DSPguru’s explanation is over 2 years old, there has been a lot of changes in the programs since that was written.

Arry, have a read of this Tute http://club.cdfreaks.com/showthread.php?t=78614 and give Sonics AC3Enc or Sonics SoftEncode a try, as these raise the audio volume during encoding to ac3, thay are just slower.

Also, unless you have the latest BeSweet installed (which is what AC3Machine uses to do any conversions), then the old ac3enc.dll used by BeSweet does not produce proper ac3 files. Only the very latest ver of ac3enc.dll (based on ffmpeg) does.


#4

One day I’ll manage to juggle enthusiasm to help along with attention to detail. Trouble is, the latter can be a slippery one! Apologies Arry, for neglecting to check the date on that thread. checks the dates again just to be sure LOL

TZ


#5

Thanks heaps chicken dude

Do I need to use a separate program to extract the sound to a wav file first before using Sonics AC3Enc. I am backing up a music dvd by the way.

Or do I simply still use DVD2SVCD to do this and use the wav or .mp2 file that is there.
If I do this do I use the do not convert audio option or do I untick this.

Also what about azid gain…I am also confused about the audio bitrate as DVD2SVCD seems to only support up to 384kbs…

I am a little confused dude.

Any help would be appreciated.

I basically want to have the audio on my copy exact as the original …:slight_smile:


#6

Hmm… now I’m confused, first your talking about AVI to DVDR conversion “convert DIVX to DVD-R I always find that the sound plays a lot softer on the copy.” and now your talking Music DVD to DVD. :frowning:

A music DVD “normally” has either 2ch or 5.1ch AC3 and many times WAV (LMPC). If AC3, then just select “do not convert audio” option and you get exactly original audio unaltered. If WAV, then this will be stereo ONLY, so just select 224 bitrate, let D2S extract it and convert to MP2 and dont make an image. Manually convert the WAV to AC3 (bitrate 224) as per the Tutorial and author the *.mpv and new *.ac3 file in TMPGEnc DVD Author as per the DivX to DVD Tute.


#7

hmmm so what you are saying that if I backup a movie and select “do not convert audio” then it will be the same as the original…so I then do not have to run it through AC3 machine like I always do?? hmmm interesting

Maybe thats why the sound is always soft as I tick the do not convert box but I then run it through AC3 machine…I thought this was necessary for compliance.

Also with the DivX to DVD tute I always UNSELECT the do not convert audio box and then manually run it through AC3 machine…hmmmm

I do this because I find half the time I am adding the AC3 file into TMPGenc DVD Author that it will not accept my AC3 file…

Look forward to your response mate :slight_smile:


#8

Carrect, “do not convert audio” means do not convert audio, therfore IDENTICAL to original. If a DVD then the stream is 100% Dolby compliant. If ac3 in an avi, then it MAY need to be FIXED by running through BeSplit/BeSliced.

Maybe thats why the sound is always soft as I tick the do not convert box but I then run it through AC3 machine…I thought this was necessary for compliance.

AC3Machine will comvert the ac3 5.1ch to stereo ONLY ac3 without downmixing all channels. Just takes front left and right, thats why its quiet.

Also with the DivX to DVD tute I always UNSELECT the do not convert audio box and then manually run it through AC3 machine…hmmmm

Since 90% of avi tend to have MP3 audio track, then what your doing is correct. I’d use latest FFMpeg instead though.

I do this because I find half the time I am adding the AC3 file into TMPGenc DVD Author that it will not accept my AC3 file…

AC3Machine links into BeSweet for its conversions. Old versions of BeSweet did NOT make 100% compliant AC3 files, the latest is very good I find.


#9

OK, I think I understand now.

So what you are saying is that all of the AVI’s that I have converted to DVD that have AC3 sound, and where I have used AC3 Machine, it is not in 5.1?

So if an AVI contains 5.1 AC3 I should use Sonic software?

So with an AVI, also what about azid gain…I am also confused about the audio bitrate as DVD2SVCD seems to only support up to 384kbs…

If the AVI is say 5.1 AC3 448 bitrate, how would you convert it so that the resulting DVD was AC3 5.1 448 bitrate…

Sorry if this is confusing dude :slight_smile:


#10

phew…

So what you are saying is that all of the AVI’s that I have converted to DVD that have AC3 sound, and where I have used AC3 Machine, it is not in 5.1?

Correct, AC3Machine can make ONLY 2ch AC3 files.

So if an AVI contains 5.1 AC3 I should use Sonic software?

No, see below.

So with an AVI, also what about azid gain…I am also confused about the audio bitrate as DVD2SVCD seems to only support up to 384kbs…

If the AVI is say 5.1 AC3 448 bitrate, how would you convert it so that the resulting DVD was AC3 5.1 448 bitrate…

As I said above, if the avi contains AC3 audio (be it 2Ch or 5.2Ch) then just tick the “do not convert audio” option and it WILL NOT convert or manipulate the audio in any way. So a AC3 5.1 448 bitrate audio track ends up as a AC3 5.1 448 bitrate. This is not rocket science here :smiley: All it does is extract it and mux it into the final VOB files, unaltered. The bitrate option is not available when you tick “do not convert audio”, so you cant select anything anyhow. Any audio bitrate setting is when the avi contains MP3 audio (90% of avi’s) and you would select either 192 or 224 ONLY ! D2S would convert this to MP2 audio and then you would run either this MP2 or the WAV through BeSweet/AC3Machine/ffMPEG/etc etc to convert to AC3 (set at the same bitrate as you selected in D2S).

Sorry if this is confusing dude :slight_smile:

Its :cool: but I suggest you re-read the tutorial and stop trying to read between the lines, theres nothing there.


#11

Thanks again…sorry If I seem to be a pain…but I was confused mainly due to your comments here from your DIVX to DVD Tute

  1. Click the Audio Tab. The Audio 1 downsample 48 -> 44.1 option will be greyed out (or absent). Again, 99% of DivX’s have 48khz audio, however if it was only 44.1khz, then see NOTE 3 below as it MUST be manually converted to 48khz later. For Audio 1 Bitrate, select 384 if the Audio was 5ch AC3 and 224 if 2ch AC3 or MP3. The box for Autodetect Azid Gain (2 pass) should be ticked as this with Normalize the MP3 audio sound level. If your AVI containes 48khz AC3 audio, then and ONLY then select the Do Not Convert Audio option. Read the warning that pops up and use at your own risk.

The parts that confused me were in Bold.

From reading it I understood that you select 384 if AC3 5ch, which is what I have been doing and then running through AC3 Machine which is the reason for my softer sounding output.

I also thought that since you said Use the DO NOT CONVERT AUDIO at your own risk that it wasn’t the way to go.

Thanks heaps anyway as I have been using your tutes for almost 2 years…the only issue ever being the softer sound. :smiley:


#12

Well glad to hear its all sorted.

The comments in the part of ther tute above “Read the warning that pops up and use at your own risk.” is for those that have a “normal” AVI that has MP3 audio, they tick the “Do Not Covert Audio” option and then wonder why the authoring crashs. :smiley: Maybe I could have spent some extra wording to explain a little more indepth but then again, I dont want to turn it into a novel, its probably to long now. Trying to satisfy the needs of the absolute biginner, the average user and advanced users in the one tute is very difficult if not impossible. :wink:


#13

There’s a quick and dirty way to deal with the occasional AVI that has low volume. AC3Filter is a DirectShow filter for decoding AC3 tracks in AVIs and DVDs. Get AC3Filter and install, it’ll assign itself first priority. When you come across a low audio AVI, go to Configuration and increase gain and check normalise. I never had a problem with gain set to +20db and 2/0 output. Not ideal, but it’s easy. (I used to play around with AVIs a good bit just to see what I could do to fix crummy DL’d ones. The more buggered and corrupt, the better. Really they’re not worth the effort, quite apart from copyright issues.) :wink:


#14

I am having the same problem arry had. I am using d2s and the CM tutorial on dvd to dvd-r: http://club.cdfreaks.com/showthread.php?t=62671

If I convert ntsc to ntsc I have no problems (do not convert audio is ticked). If I convert PAL to ntsc, then my AC3 sound level comes out low. I use the latest besweet (1.5 b29) to do the ac3->ac3 conversion with pal to ntsc time streching done (step 10 of the pal->ntsc conversion).

So my questions are:

  1. Should the volume level of the time stretched ac3 differ from the extracted ac3? It seems to me that the volume of the stretched ac3 is softer…

  2. Is there a way in besweet to increase the volume of the ac3 as I am doing the pal to ntsc conversion using besweet?

Here is what I have been doing. I have been using headac3he to extract the .ac3 to a .wav (using the heavy option on the ac3 stream to increase the volume). Then I use goldwave to maximize the .wav. Then I use besweet to time stretch the .wav and create a .mp2. Then I reauthor the DVD.

Why go through all that you ask… I read a great tutorial on AC3 encoding on doom9: http://forum.doom9.org/showthread.php?s=&threadid=56020
and it scares me. I have no method of determing the “rms level of a .wav file” so I can not determine the “dialog normalization” parameter. Therefore I can not create a proper ac3 file. And when I try and guess at the levels to use for dialog normalization the ac3 turns out REALLY bad…

  1. Is there a more concise method to take an ancoded ac3, raise the volume and still have an ac3 file? It seems to me that if I tell d2s to extract the audio, then I will have a normalized .wav file that I can send through besweet to do the pal to ntsc conversion and then I can use that wave and ac3enc (as per CMs guide) to create a properly encoded ac3. However, CMs guide says not to mess with the dialog normalization parameters in ac3enc and the doom9 guide says you have to set the dialog normalization parameter by knowing the average RMS level of the .wav…

  2. If I did use the d2s outputted, normalized .wav file to create the ntsc source AND I could measure the average rms level so that I could set the dialog normalization parameter, would the ac3enc file turn out louder?

Thanks for any replies…


#15

Why dont you spend a few minutes and tryout ac3enc, it has always worked for me when I have low sound levels. BeSweet has a know bug/feature of lowering audio abit dealing with AC3, thats why I never use it for such.

Besides, have a read up on DGPulldown, as using that you dont need to stretch/shrink the audio when doing either PAL to NTSC or NTSC to PAL. This then eliminates the BeSweet step. Its at http://neuron2.net/dgpulldown/dgpulldown.html Maybe if I ever get the time (I have really retired from here :smiley: ) I will update the Tute. Maybe even when a new version of D2S comes out :slight_smile: Just maybe.


#16

Thanks for the reply CM.

Here is the experiment I ran last night. Taking the extracted .ac3 file from d2s I used headac3he .23 to create a .wav (used output heavy option to increase the dynamic range). Then I used besweet to convert the 25fps .wav to a 23.976 .wav. Then I used cooledit pro 2.1 to analyze the .wav and find the average rms. When I ran ac3enc, I was able to set a valid value of dialog normalization (-20db ibstead of default -31db) and I encoded. Once I reauthored the dvd, the soundtrack was in good shape.

I have started reading up on dgpulldown. Reading through that 10 pages of posting on the subject is grueling. I consider myself an intermediate user. I learn from reading that stuff, but much of it goes over my head.

Wow… It seems to good to be true. The link CM provides to the dgpulldown states the work flow is:

  1. Demux the video and audio streams.
  2. If the video is interlaced, deinterlace it to make it progressive. If it is already progressive, skip this step.
  3. Resize the video to 720x480.
  4. Encode the video to MPEG2 at 25fps progressive.

It seems to me that if I follow the dvd-dvd guide that CM made, along with the pal->ntsc conversion guide at the end of the post i.e. edit the script before serving and simply change the 576 to a 480 on the resize (do not add the assumefps 23.976), save and execute that the above 4 steps will be done by d2s. Am I correct in that assumption?

  1. Run DGPulldown on the video MPEG2 stream to flag it up to 29.97fps (using the 25fps --> 29.97fps conversion).

Run latest version of dgdecode with 26 -> 29.97 checked.

  1. Mux the audio and video into a program stream, or author a DVD using the audio and video streams.

Using dvdlab.

Is it that simple now to do the PAL -> NTSC conversion?

Thanks for the pointers, CM! I really appreciate it.


#17

Yes it is.

Just simply encode the PAL to PAL but set the pic size to be 720x480 instead of the normal 720x576. Tick the “Edit as Part of Video Encoding” option in Frameserver and edit window when it pops up or create a Special Avisynth setup by editing the avisynth.ini file by adding something like :

[AVISYNTH_LanczosPALtoNTSC]
0=Import(!RoBaConditional.avs)
1=LanczosResize(720,480)
!RoBaConditional.avs=E:\DVD2SVCD\Tylo\RoBaConditional.avs

If any borders need to be added, just tick the “Edit as Part of Video Encoding” option and manually add them. The avirez program can help with making the numbers easy for you :slight_smile:

Dont make any images, of course, and simply run the created Encoded_Video_CCE_PAL.mpv through DGPulldown using the 25.000 --> 29.970 option. Then author in any program of your choice :slight_smile:

EDIT: Forgot to add, if the PAL source is interlaced, then just select the Telecine (PAL) option in the Deinterlace box.


#18

Ugh… I had a feeling that interlaced question was going to show up. I guess I have to do some searching. I have no idea how to tell if the source is interlaced. Is there an easy way to tell? Is that Telecine (PAL) option in the d2s program? Guess I have never had to deal with that before.

I use avirez when converting pal .avi to dvd (again using your wonderful guide). I also use gspot to get the .avi resolution to feed into avirez.

Is there an equivalent to gpsot to use on vobs? As I recall d2s is pretty smart when doing the dvd to dvd processing (as opposed to the .avi to dvd in which it is pretty braindead). In dvd mode it automaticlly selectes 4:3, anamorphic, or the encode widescreen as fullscreen. When the avisynth script pops up on these the resize simply is 720, 576. I have just been putting a 480 in there. Is it same to assume that if borders have to be added that d2s will generate the command in the avisynth script? If they are in there I then have a queue that I need to change them.

Thanks again. I am going to try this processing this weekend.


#19

Yes… the Telecine option is in D2S :smiley: I use Bitrate Viewer (www.tecoltd.com) to tell me if its interlaced or not. Load a VOB file in and look at what it says for DCT Type. If FRAME, then its Progressive (not interlaced), if it says FIELD, then its interlaced. Ignor what it says for Frame Type, its not always correct. Most PAL stuff is interlaced.

I use avirez when converting pal .avi to dvd (again using your wonderful guide). I also use gspot to get the .avi resolution to feed into avirez.
Is there an equivalent to gpsot to use on vobs?

Yes, use GSPOT ver 2.51 Beta, works well with AVI’s, MPEG’s & VOB’s.

As I recall d2s is pretty smart when doing the dvd to dvd processing (as opposed to the .avi to dvd in which it is pretty braindead). In dvd mode it automaticlly selectes 4:3, anamorphic, or the encode widescreen as fullscreen. When the avisynth script pops up on these the resize simply is 720, 576. I have just been putting a 480 in there. Is it same to assume that if borders have to be added that d2s will generate the command in the avisynth script? If they are in there I then have a queue that I need to change them.

For DVD’s there will never be borders added if going from 4:3 to 4:3 or 16:9 to 16:9 (they are the only 2 formats they come in anyway), and as you said, D2S gets that right all the time. Its just with AVI’s, as they can have any pic dimmensions and not anamorphic. If doing a PAL to NTSC AVI, then use AVIRES to calculate the borders for an NTSC film.


#20

CM- As always, I appreciate your insight. Well… Here I go!