Another thread about Normalizing

[Russell Landwehr]

Speaking of genre dependent, I just did a bunch of "drone" cues. I "mastered" the cues lightly using a multi-band compressor and an L2. BUT, I also ended up mastering them at about -3.0 or -4.0. And I didn't normalize.

Here's what my thinking was on this. "Drone" cues are suppose to be subtle. I figure an editor is more likely to use a cue that sounds more subtle right out if the gate than one he's got to turn down because "whoa! THAT'S not subtle."

Russell

[Len911]

The downloadable tracks I got were mp3 128bit

One process I was going to do, well I actually did it, ear frequency compensation normalizing in wave lab, however part of the algorithm of an mp3 is that very thing.

Maybe that's the topic of another thread. How effective is effecting mp3's? Personally, I can't tell that anything I did made that much difference other than gain changing.

[andygabrys]

Since you asked, and I have opinions on everything

This....

Logic Pro X
-->Tracks --> Output --> Master out

is the ONLY thing that affects your actual production. How hot stuff is recorded (if its live) or hot hot something is coming out of a VI plugin > how hot the submix busses are pushed > and hot much actual volume is hitting the 2 bus / output / master out.

There is a set of "rules" or maybe "best practices":

digital 0.0 dB is the onset of nasty digital clipping. It can't be avoided. Stuff that peaks just below digital 0.0 dB will be fine. Using a limiter set with an output ceiling of -0.2 dB for example in most cases produces clean bounces to AIF / WAV or to mp3. How hard you drive the tracks into that limiter, or how much gain you add in the limiter itself governs how much dynamic range the resulting production is going to have.

In most cases a dynamic range of 20 dB (also called crest factor) is found on cinematic material.
a dynamic range of 14 dB is hot enough for most productions.
a dynamic range of 12 dB is hotter still and will cover or exceed most broadcast specs
most commercial music produced today has a dynamic range of between 4 dB and 10 dB. Hybrid styles, metal, and electronica are often found with extremely low dynamic range.

Whether or not you agree with these specs, or like the music produced under these conditions is a whole different thing.


This....

Macbook Air system volume --> Elgato Thunderbolt Dock
Elgato Headphone output splitter -->
--> Sony 7605 headphones
--> Behringer 802 4/6 channel mixer
--> AKG K240 headphones
--> Samson Resolve50a nearfield monitors (each has its own volume knob)

only affects how it appears in your room. How loud it seems coming out of your headphones, your monitors etc.

You can have a "quiet" production with a lot of headroom below digital 0.0, and it your monitors are turned up way loud, it will sound loud and impressive.

or you could have your monitors turned down as far as they can go, and you can play a commercial piece with a dynamic range of 5 dB, and it will sound quiet. Because its been limited so much, you might hear every thing in the track, but it won't be "loud". Until you turn up your monitors.

[mojobone]

Andy, you should talk to the guys (and presumably gals) that master death metal; it's pretty routine in those circles to clip at the converters. Just don't ask me why.

Very simply put, normalization is a digital process, usually destructive, (but you saved a copy, beforehand, right?) that increases the gain of a digital audio file to a user-determined level relative to absolute zero (0dBFS) WITHOUT disturbing the relationship of audio peaks to RMS average. Usually, there's one knob that sets the level of the largest peak in the file relative to absolute zero (no headroom at all)


I've been at this game a while, but I only just today figured out why my DAW's master fader defaults to -3dBFS when every other fader defaults to nominal. -3dBFS is exactly half the energy of 0dBFS, so if you're doing parallel processing, it eliminates a step. It also gives you three entire decibels of headroom to play with before you start compressing.

[Len911]

This thread sort of ebbs and flows between normalization and normalization-not

It's basically a volume control. If you have a volume knob with 10 settings, and mostly you listen around 3 which is what it sits on when you listen to the world, and now you have an audio file that you must turn up to 10 to achieve the same volume, you might want to normalize it, so you can listen at 3 like the majority of your listening.

What it isn't going to do is change the dynamics. So if you have a gunshot and a pin drop in the audio and need to hear the pin drop, you will have to use compression to lower the dynamics so you can fit in the pin drop. You are effectively turning down the gunshot, so the volume can be increased and the pin drop heard. Because of the extremes, the dynamic range would be almost non-existent.

If you wanted to set the ideal dynamic range, you could copy the quietest note to a track, and copy the loudest note to the track, and set a compressor/limiter so that the quietest note was heard at the level you desired, and the loudest was reigned in and lowered to maximum before distortion, 0db theoretically max., however, these are ballbark settings unless you take everything and after all processing
at the same point in time, because it all adds to the gain. If that wasn't contradictory,lol, but in reality it's theory that might help in understanding the process, because nobody is actually going to do that, I don't.

Wave Lab has a "Loudness Normalizer". It has a "peak limiter" associated with it if needed. It's a little different animal,lol, and it also has an option for "ear frequency sensitivity compensation". It applies the Fletcher-Munson curve if you will, that the human ear is more sensitive to freq in the vocal range or mid-range than the low or high freq. And that is sort of what happens in the mp3 algorithm. It's interesting. I've never had the peak limiter to be needed. I've only experimented with it. There tends to be ranges for certain types of material, -27 to -20db is probably as low as you wanna go with orchestral type tracks, rock maybe down to -13 or -14db, you can sort of find the sweet spot, edit>undo and try different loudness settings. It's quick,lol!

[andygabrys]

Yeah Mojo I have read a lot of articles on that kind of mastering, more just trying to understand how and why more than anything.

3 things I learned.

You have to take stuff into the analog domain (duh!) cause you have to have converters to clip,

There seems to be a preference for the clipping at the convertor stage due to sound preference (an article I found just the other day which I will try and find)

Most of these mastering people doing this process seem to almost unanimously using Lavry Blue or Gold converters for the A/D stage.

[mojobone]

As I live and learn, there's apparently a fan for every flavor of distortion, and not for nuthin', but Burl makes A/D converters that are designed to give you analog flava when run hot. Besides, I'm at least nominally still a guitar player, soooo....i can haz no right to biatch about someone else's distortions.

[sansharbour]

Great thread

Paulie Keep the faders at 0
and as Andy said calibrate your monitors. It's not that hard.
When I first got my DAW i carried over the Get it close to the Red mentality from my porta studio days, which did in fact result in some digital distortion (not good)
Then I started reading and using Lynda.com training on Logic Pro x
Around that time I discovered Ian Shepherd and his videos and blogs on the topic of Dynamic Range and the Loudness Wars

It made a whole lot of sense to me. He uses meters and I bought some too and i love the things.
I think I am a tad conservative in the volume dept when compared to others
for example I usually put the final limiter at .02
hmm maybe I should raise it......or send a hotter signal to it as Andy said
I'm still learning

Matt you say you master hot. Is that 0.0 , 0.05, 0.1

You can be as loud as you like on a cd but not on ITunes because they are going to lower it
One thing you can see in iTunes is how much they will bring your song down in db if you look at the file in the get info section . Look halfway down to the volume box.
Here you will see how much they have lowered your song to make it equal.

Charlie X Boom Clap brought down by -!0db
Paul McCartney Peace in the Neighbourhood lowered by -2 db
Keith Urban Good Thing lowered by - 7.9
There are new standards on iTunes and You Tube

So I guess in the end, Volume is a choice we have to make and if it sounds good to you then that's what counts
A lot of the Taxi submissions I hear sound pretty darn good

Good Luck
Don

[mojobone]

In most cases a dynamic range of 20 dB (also called crest factor) is found on cinematic material.

Just a quick correction: The difference between the loudest and least loud points in an audio file is the dynamic range, the crest factor is the difference between the peak and the RMS average at a given point in the file.

[andygabrys]

In most cases a dynamic range of 20 dB (also called crest factor) is found on cinematic material.

Just a quick correction: The difference between the loudest and least loud points in an audio file is the dynamic range, the crest factor is the difference between the peak and the RMS average at a given point in the file.

thanks Mojo my bad.

my misuse of the term dynamic range is the issue.

what I should have said is:

cinematic material might have a crest factor of 20 dB

etc.