Stereo Tool

A continuation of the discussion, initially between myself and Wes Keene, beginning hereish:
viewtopic.php?f=14&t=7889&start=320#p36722

---
It's only relevant IF presets are calibrated to something, otherwise you DO need to adjust the pre-amp gain for each preset. The AGC doesn't work based on peaks, and of course it should not. Nor should anything else that's influencing the final loudness. There's no point in normalizing playback to a maximum peak, when the track-to-track loudness can have such a huge disparity that you need to use aggressive AGC settings with huge ranges anyways, making normalization a moot point. That also makes adjustment of the pre-amp a moot point, since you can't set the AGC gates anywhere near something optimal. The gates don't work based on peaks, which is also the correct way to handle that. With peak normalization, the point at which the gates start working will be different from track-to-track.
I agree, except that peak normalization isn't the way to go. It doesn't solve any problems. Very rarely are digital recordings not hitting 0dBFS somewhere anyways, so it doesn't do much.

But like I was saying before... if someone hasn't used loudness normalization & metering before, and heard how much better the overall presentation can be... then they'll be used to preferring processing that deals with it by either -A- trending toward an AGC with infinite ratio and high speeds and range. -B- Without the ability to attain that, they'll trend toward minimizing the track-to-track inconsistencies by smashing everything at every step as much as they can stand. Bye bye TSL.

Obviously "-A-" is preferable for the listeners, but it removes all macro dynamics.

If someone is not using loudness normalization, and is designing presets for other people that don't use infinite ratio for the AGC, then I question their abilities, and want to help educate them on why this makes the problem of peak normalization worse when it comes to audio processing. That discussion touches on a lot of interesting inter-related subjects, but I digress here. (I still haven't looked at your preset Wes, so this isn't directed at you, it's just my opinion from a technical & philosophical standpoint)
We're engineers. We misunderstand people as much as we're misunderstood. All good in the hood.
-12dBFS would be, yes. Even with peak normalized music, if you have mics, they will be mixed at least at the same loudness as the music. Then you already have at least less than 6dB headroom to 0dBFS with just one mic (if you have a VO with extremely wide spectrum, pre-processed with a multiband hard before the final audio processor, with the same density as whatever music is playing). Most VO are more dynamic, thankfully, which means even less headroom.

Many people ride the faders at least little bit on the mic, or use auto ducking. I prefer subtle auto ducking myself. So that headroom isn't quite as low, but the potential is there.

With loudness normalization and loudness metered mixing at 0LU, there's more than enough headroom for almost anything, and no need to even thing about what the peaks are doing. That's why R128 (specifically EBU tech 3341 "‘EBU Mode’ metering to supplement Loudness normalisation") meters don't even show the peaks. They do require a clipping indication, at -1TP, but they don't want people to be influenced by even knowing what the peaks are. It's about freeing people up to be able to make the program material sound however they want.