Audio normalisation when backing up concert DVDs to MP3

[osullic]

At the risk of asking the obvious, I noticed that one of the steps that Auto Gordian Knot performs is to normalise the audio, and I was wondering why this is necessary.

Is it the case that audio encoded to commercial DVDs is somehow lower in volume than "normal"? Or is it just always good practice to perform audio normalisation?

What I have in mind here is several concert DVDs that I'd like to back up, both to AVI and to MP3. To get the MP3 audio, I was just going to retrieve the MP3 file from the agk_tmp folder that Auto Gordian Knot creates, and I'm just wondering if it might not be better to do it some other way in order to avoid the audio getting normalised. For comparison, when ripping CDs to MP3 with Exact Audio Copy and LAME, I don't perform any audio normalisation, so I'm wondering why/if I shouldn't also skip normalisation in the case of extracting audio from a DVD.

Sorry if I'm not understanding the process properly.

[smok3]

Perhaps read on this https://tech.ebu.ch/loudness to get to the core of slippery named "normalization".

[hello_hello]

You can't stop AutoGK from "normalising" if it converts the audio, but if you elect to keep the original audio instead (AC3 etc), then obviously it won't. You could convert the AC3 to MP3 via other means if you don't want to normalise. For the record, AutoGK also downmixes multi channel audio to stereo using Prologic. I can't say I've ever tried to decode the stereo/Prologic audio to multi channel again, so I've no idea how well it works.

I don't like volume normalisation as a rule. Well for movies it's probably no big deal but for the audio of episodic video (multiple episodes on a DVD for example), normalising can adjust the volume of each slightly differently, or sometimes very differently, so I avoid normalising those.

If I'm converting stereo audio to stereo I leave the volume as it is. For downmixing 5.1ch to stereo...... well I don't do much of that these days given I don't encode with AutoGK any more (MeGUI and the x264 encoder instead).... but if I do downmix to stereo I re-encode while reducing the volume by 6dB to prevent any clipping which might result from the channels being combined (6dB seems to be enough). That way even if the volume changes a little as a result of downmixing, it changes by the same amount for every file I encode, so the relative volumes should remain unchanged.

I tend to do pretty much all of my audio encoding with foobar2000. It has a dsp for donwmixing to stereo which can be used when re-encoding, and you can adjust the volume manually. Foobar2000 doesn't normalise.

PS. Some programs (I'm not sure about AutoGK) downmix multichannel audio using a matrix which reduces the volume quite a bit (it's the prevent clipping thing again). If that's the case, normalising might be an idea, otherwise the volume might be very low.