Volume levelling of audio tracks

[DonR]

Hi all,

Wasn't sure if I should pitch this in the DVD authoring forum or here but seeing as it's purely an audio question I'll post here. I'm putting together a home movie DVD which has background music for the menus, music track for the slide show and audio for the movie itself naturally. All the audio has been converted to AC3 using BeLight from either a WAV or mp3 source. I've added them to my DVD authoring with DVD Lab Pro and now have a DVD with varying degrees of audio volume across the different elements. I was wondering if there is a tool where all the levels can be equalised, normalised? Thanks,

Rgds,

Don

[raquete]

is one good question;) but how much is the ideal level? (i ask together with you)
right now,i'm reading this big articles that i found in another forum:
http://www.digido.com/portal/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=36/
http://www.digido.com/portal/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=59/

as you are treating AC3 and if you're interested,i did some samples exactly to ask and find the "ideal" level(if it really exist) :
http://forum.doom9.org/showthread.php?t=114390
download the VIDEO_TS folder that have all samples inside.

for "fast solution" to normalize,audio editors like soundforge, a.audition,wavelab(and others) can do that easily with .waves

regards.

[Skelsgard]

BeLight for AC3 encoding is not the best option regarding quality. If u can, compress the audio again using Aften for AC3 encoding.
As for a tool for normalizing audio in batch mode, Belight itself is a great tool.
Just set the Normalize levels in the Advanced Settings tab.

http://img19.imageshack.us/img19/5827/belightro6.jpg

Wavewizard can be used too.

http://img212.imageshack.us/img212/1084/ww2ed1.jpg

Cya.

[DonR]

Thank you for your replies. I'll give it ago and see (or hear :)) how it goes!

Rgds,

Don

[loubat]

Hate to hijack this thread, but I found it while searching, and it seems to be the most relavent topic to my problem. I've got 5 videos that I want to put on a dvd. One of the vids had a much higher volume than the others. So I'm trying to use BeLight to lower the volume of the track. At first I couldn't get BeLight to do anything at all, but I downloaded it from another site, reinstalled it and now it seems like it's working, at least it says it is.

First I open the file (an .mpa file), click the mp2 tab, click Advanced, and do normalize to 70%. I run it though and once it's done processing the file, I open up the new file in WinAmp and compare it to the original and they are exactly the same volume, no change.

So then I thought, maybe normalize is just bringing the highs and lows to a more middle of the road level, not actually *lowering* the volume by 70%. So I go in and try to set a -10db gain. Run it through again, and still no change in volume. Is there an easier way to change the volume of an audio track, or am I just screwing something up?

[ThaiJan]

As 'loubat' I excuse for continuing the hijack of this thread, but I got exactly same problem as mentioned by both 'DonR' and 'loubat'. BeLight won't do the trick for me either. Can anybody help or is the silence meant as a towel in the ring?

Cheers

[DSP8000]

Just get
wavegain-1.2.6 (http://www.rarewares.org/files/others/wavegain-1.2.6.zip) use the GUI from this zip file but replace wavegain with 1.2.7b wavegain-1.2.7beta (http://www.rarewares.org/files/others/wavegain-1.2.7beta.zip)
Use track/radio gain for single track.
IMHO this is the *best way to 'normalize' :).

DSP8000

[ursamtl]

I agree with DSP8000 wholeheartedly! Wavegain processes the audio using something called Replaygain, which is far more effective and potentially less damaging than "normalizing." Humans perceive loudness based on the average level (RMS) of audio not by the peaks. All normalizing does is adjust the audio so that the peaks reach a certain level. It finds the loudest point of a file and sets it to the target. If you have a file with one or a few loud transients but the average level is low, then even after normalization, it is not going to sound loud enough. Plus, if you normalize to a level that's too close to digital 0dB (100%), chances are that you'll suffer from overshoots that clip off part of the soundwave. As Bob Katz points out in articles (I believe Raquete linked to them. They're definitely on the digido site), many electronic components start distorting signals before they reach their clipping point, so if one is going to normalize, it's a good idea not to go higher than -3dB (70%).

Replaygain works by measuring the average level and adjusting it to preset levels that are based on scientific studies of comfortable listening levels. These levels can of course be changed to match personal preferences.

In addition to wavegain, Replaygain is built into Foobar2000. I use it regularly and find it helps a lot, especially with all the overcompressed, maxed out releases that are happening lately.

For more information, www.replaygain.org is a good starting point.

Regards,
Steve.

[Mug Funky]

it might be worth mentioning that the audio should be re-encoded from the original sources (rather than decode ac3, wavgain, re-encode).

there's a batch on the last page of the Aften thread that automatically scans volume, then inputs the dialog level found into aften to encode with. this will (hopefully :)) give you perfectly consistent audio levels.

perhaps some time in the future we'll have a utility that scans an ac3, and re-writes the dialnorm and DRC data without touching the actual audio data - that way audio quality is not compromised at all.

[ursamtl]

it might be worth mentioning that the audio should be re-encoded from the original sources (rather than decode ac3, wavgain, re-encode).

there's a batch on the last page of the Aften thread that automatically scans volume, then inputs the dialog level found into aften to encode with. this will (hopefully :)) give you perfectly consistent audio levels.

perhaps some time in the future we'll have a utility that scans an ac3, and re-writes the dialnorm and DRC data without touching the actual audio data - that way audio quality is not compromised at all.

This is a really good point. Trying to do any processing on a file once it's been processed with lossy compression can produce really bad results. Your proposed utility is a very solid idea. Perhaps an implementation something like Replaygain would be good, where it's simply a value that the host app reads and then adjusts its level accordingly.

[ThaiJan]

Okay, I was pretty sure you knew the answer :D .

Sounds a little technical for a beginner, but I'll get on with it.

Thanks again.