Normalization Question

[Mjespo125]

Allright guys, I don't know if this is a stupid question or not, as this isn't something I've done maybe I'm being mislead. I'm at college studying audio production and one of my fellow students told me that I should normalize EVERYTHING I record to optimum levels? Is this common practice?, and what do you guys normalize? I never normalize because I am afraid that it introduces distortion to the tracks, but maybe I'm wrong?

 

[bryan_kilco]

I'm pretty sure you don't want to do that. You want your mix to be quiet. Loudness comes from Mastering.

 

[koalamo]

nooooo

 

[Mjespo125]

I figured that this dude didn't know what he was talking about, but normalization must have its place somewhere right? Where would you use it?

 

[bryan_kilco]

Someone correct me if I'm wrong, but Normalizing basically is making the waveform as fat as it can be without clipping? Sort of limiting, in a way?

 

[MindMunch]

Normalization is when you raise the volume of the entire signal until the highest peak reach 0db. It´s basically useless when tracking, mixing and even mastering, as you don´t want it to really reach 0db. The only scenario where I ever felt the need to use it was when I wanted to put the shotgun gunshot sample from Doom on my cellphone as the message ring. I just normalized the sample and reduced 1db.

 

[Skinny Viking]

smack your buddy in the mouth

 

[DestroyTheColossus]

I would never normalize for music. There just isn't a reason to do it, it will take up your headroom for mixing, and if 2 things are supposed to be balanced in volume (such as guitar tracks, overheads, any single track with multiple events, etc.) it will mess that up too. I would only use it when making 1 off sound effects and even then its not a necessity.

 

[Peludismo]

There are 2 ways to normalize, one is to peak level (the one that mindmunch named) that takes the highest peak to 0 dB (or any other value that you want), and the other ways to normalize is to RMS, in other words, to compress. Sony Sound Forge has this function that you simply say "i want the RMS to reach -13 dB" and the soft does it. The only cool thing that SF has is the function to normalize the RMS considering the equal loudness contour, the Fletcher-Munson curves.

The only case i would use SF to Master is if i really am in a hurry and have to deliver some files and don't have the time to really listen how the limiter is working, etc. But i don't like SF

Sorry for my english, i'm not used to write all the time, so my way to speak is kinda odd

 

[Arsenu,]

your buddy sounds like i did after about 2 classes of cubase....

nowadays i don't normalize AT ALL. the main reason bieng that the normalize process just picks the loudest peak (could be a 1 ms peak) and raises it to 0db and the raises everyhing else at the same ratio, which means that if you have the slightest glitch which louder than everything else the signal will be normalized according to that glitch, anyway i lost my train of thought but generally normalizing is BAD. use a c1 compressor and that's it....

 

[BLUElightCory]

Just in case the point hasn't been driven home, your fellow student is wrong, wrong, frigging wrong.

 

[timislegend]

i think that maybe his terminology was incorrect. maybe he was thinking that to "normalize" audio is to "even-out" or matching one audio source with another.

it's pretty common practice to use normalization however (like everything in digital audio) it depends on what you are using it for.


if he said... "select every region and normalize them..." ??? well, then that is destructive and probably not the best thing to do !!!although!!! i have seen some engineers use normalization for some really ridiculous things. =\