AC3 Dialogue Normalization so it doesn't change volume of edited original

[c627627]

If I wish to slightly edit an ac3 file, whenever I save it in Sony Sound Forge, the resulting volume is always lower.

The AC3 Dialogue Normalization is set to default -27dB. The options are from -1db (lowest volume) to -31dB (highest volume.)



What should I set it to so that the volume of saved file is identical to the original? It appears that when I set it to max -31dB that I get the volume level closest to the original.

[nic2k4]

Actually, it's the other way around; 0 dB is the highest level you can get, anything above and the waveform becomes a flat line (clipping).

[c627627]

Sony Sound Forge > File > Save As... > AC-3 Pro > Custom... > Audio Service TAB > Dialog normalization: >


Now choices are from -1 dB to -31 dB and the resulting file is the quietest if -1 dB is chosen and loudest if -31 dB is chosen, I just retested it.



What this is about: When you open an mp3 or a wav file and you move things around and you save, there are no changes to the audible volume of the file.

Why are there changes to the volume of the slightly edited ac3 file? All I want is for volume to stay the same after I move a few things around inside the file. I don't want any "normalization" to take place. Why does it "have to" for ac3 files and not for mp3 or wav files?

[nic2k4]

As something to do with the mixing of all the surround channels, don't know the details I usually downconvert to 2 channels and have totally different issues. Last time I used soundforge it was still a sonic foundry product; I'm not sure what their level settings mean, but if -31 is the same as 0 dB for them, that's how it is. All I can say is if I'd want to keep the level the same, I would set it to 0 dB i.e. no attenuation (-31 for you).

[Richardm]

If you don’t want any changes to the volume set it to -31dB.
Here is good explanation how Dialogue Normalization works:
http://www.hometheaterhifi.com/volume_7_2/feature-article-dialog-normalization-6-2000.html

[c627627]

Thank you for that link. Let me know if I am correct:


1. To quickly move around parts of the ac3 file in order to replace the original, set the Dialog normalization value in Sound Forge to (the highest available) -31 dB value.


2. But if you have captured audio from say, a TV in wave format, then it's better not to mess with that Dialog normalization value and just leave it at its default -27 dB value when saving that wave file into ac3 when making a DVD, is that correct?

[geminigod]

I know this is an old thread, but since it popped up as such a high google result in a related search I was doing and the information here is so misleading or wrong, I felt compelled to correct all this nonsense for future people that wind up here with the same question that c627627 had. Here is a link to a detailed explanation of the concepts and related issues as well as links to other good sources for continuing education:

http://www.fanedit.org/forums/showthread.php?4989-5.1-surround-sound-dialog-normalization

[aedipuss]

dialnorm is just the relative output of the center channel to the left and right channels. the bigger the dialnorm -db, the more reduction is applied to the center channel as compared to the output of the others.

[c627627]

So to answer your original question c627627, it really depends on what is specifically happening in the sound file you are listening to in terms of whether it sounds louder or quieter at different dialogue normalization levels, and it depends on how your sound system is setup that you are listening on.
[...]
When it comes to editing, you should never edit in AC3. Convert to a lossless format like wave, w64, or PCM. Then you can convert back to ac3 with whatever dialogue normalization level you want! Hope this helps clear up some confusion for people.

geminigod,
I was editing an Iron Maiden music DVD menu and naturally I wanted it to sound just as loud as the original. With each edit, as I recall, the overall menu volume sounded quieter. It did not sound as loud as the original.

This was not about editing a movie where the intricacies of footsteps vs. oncoming train were of concern. I wanted the menu to sound as loud as the original. I only made slight modifications, and did not wish the perceived level of volume to be different.


1. Edited menu sounded quieter than the original when Dialog normalization was set to -27 dB.
2. Edited menu sounded about the same as the original when Dialog normalization was set to the most extreme value of -31 dB.

[geminigod]

Aedipuss is simultaneously right and wrong. By coincidence because dialogue tends to go mostly through the center speaker, you will notice the most substantial changes there on a 5.1 film audio track, but dialogue normalization still very much applies to stereo audio tracks. Don't feel bad, Aedipuss, my original writeup here was also wrong in a number of ways, which is why I have edited my previous post.

c627627, it sounds like you did the right thing with your dialogue normalization setting, assuming you changed the dynamic range compression settings to none. Music is a tough call. For a music video I think I might be inclined to use -31 DN and DRC set to none, which will yield an identical sound to a wave file of the same audio, but it would be good to analyze the audio first and make sure no clipping is going on. You may find that using -27 dB, which will sound slightly quieter, is not clipping your high range, which you can then compensate for by turning up the volume! But for film I would not recommend any of those settings. (see link in my previous post).

[c627627]

geminigod, thanks for posting. What you are saying makes sense.


And by the way, yes, using Sony Sound Forge, I wrote this for myself about dynamic range compression settings:
Note that when rendering to AC3 Dolby Digital audio format, the volume will be automatically reduced.
To prevent volume reduction, go to PreProcessing TAB and change both the Line mode profile and RF mode profile to NONE. Also set Audio Service TAB > Dialog normalization: to max setting of -31 dB to avoid reduction of volume.

[geminigod]

Even with a music video those settings being the best are highly subjective to the source material, but it will probably work out mostly fine in general. It will sound loud at a low volume, which is usually what the average person wants, but you do run a risk of losing some of the music to clipping. Just to be safe, my advice would probably be to use the default of -27dB. This will be a slightly lower sound level but hopefully eliminate most clipping problems and produce a fuller sound if you just turn up the volume a little bit.

For any sort of dramatic film production though, those settings are highly inadvisable. I go into great detail in a guide about how to properly set the dolby digital decoding for this medium on the fanediting site that I linked to above. (You might have to register though, in order to access the forum there.) Sorry .

[c627627]

Would you detect if any clipping problems occurred by opening the edited file and by looking to see if the uppermost "lines" have been cut anywhere or is there another method?

[geminigod]

A decent audio editor can at the least display playback levels and track peak levels. In Sound Forge I believe the mix level numbers for each channel turn red when levels hit 0 or greater. Many audio editors also have levels tools that can just scan the track to report overall average and peak levels. So if it reported that your peak was +3.0 dB, then you should either drop the levels in your editor by a minimum of 3.0 dB before encoding to ac3 at a dialogue normalization of -31dB. Alternatively you could leave the track as-is and encode with dialogue normalize at -28 dB.