Compression vs. Normalization
Old 16th July 2011
  #1
Gear interested
 

Thread Starter
Compression vs. Normalization

What is the difference between compression and normalization and can you give examples of how they would be applied differently in the mastering process?
Old 16th July 2011
  #2
Lives for gear
 

It seems like you need Google and a bit of extrapolation
Old 16th July 2011
  #3
Lives for gear
 
Remeniz's Avatar
 

To me Normalization is just like upping the fader in a DAW software: It's just a matter of increasing the volume of the audio with no detrimental side effect - as long as you don't increase the audio beyond 0dBFS.

Compression? Compression reduces the dynamic range of audio; the difference between the quiet parts and peaks.

Although here, Google is your best friend.
Old 16th July 2011
  #4
Gear Head
 

Quote:
Originally Posted by Remeniz View Post
To me Normalization is just like upping the fader in a DAW software: It's just a matter of increasing the volume of the audio with no detrimental side effect - as long as you don't increase the audio beyond 0dBFS.

Compression? Compression reduces the dynamic range of audio; the difference between the quiet parts and peaks.

Although here, Google is your best friend.
With your explanation we dont need google. Very clear explanation.
Old 18th July 2011
  #5
Gear addict
 

Quote:
Originally Posted by errollem View Post
With your explanation we dont need google. Very clear explanation.
That's good. Although, the terms get mixed up often, and you'll see "normalization" tools or plugins that actually do compression. And bad-sounding compression at that.
Old 18th July 2011
  #6
Lives for gear
 
Adam Dempsey's Avatar
 

Verified Member
Quote:
Originally Posted by Remeniz View Post
To me Normalization is just like upping the fader in a DAW software: It's just a matter of increasing the volume of the audio with no detrimental side effect - as long as you don't increase the audio beyond 0dBFS.

Compression? Compression reduces the dynamic range of audio; the difference between the quiet parts and peaks.
Points for brevity?
Normalization = gain (to a determined point).
Compression = gain reduction (then sometimes gain).

The former is done mostly needlessly, for level. The latter is done for the sound of it, ideally.

Quote:
Although here, Google is your best friend.
Not necessarily.
Old 18th July 2011
  #7
Gear Head
 

and the winner is.............

Sorry guys, Remenez is the winner in explaining the difference of normalization and compression for dummy's, grandfathers and some masteringengineers......
Old 20th July 2011
  #8
Quote:
Originally Posted by Remeniz View Post
To me Normalization is just like upping the fader in a DAW software: It's just a matter of increasing the volume of the audio with no detrimental side effect - as long as you don't increase the audio beyond 0dBFS.

Compression? Compression reduces the dynamic range of audio; the difference between the quiet parts and peaks.

Although here, Google is your best friend.
Normalization adds an unnecessary layer of unwanted DSP to your audio file, with all the rounding errors & quantization distortions that implies.
Just say "no" to normalization & use the gain pot instead.

Also, "compression controls the dynamic range" might be a better description, perhaps?
Old 21st July 2011
  #9
Gear interested
 

Quote:
Originally Posted by neilwilkes View Post
Normalization adds an unnecessary layer of unwanted DSP to your audio file, with all the rounding errors & quantization distortions that implies.
Just say "no" to normalization & use the gain pot instead.

Also, "compression controls the dynamic range" might be a better description, perhaps?
Doesn't using the gain pot also add a layer of dsp? i'm genuinely puzzled why one is better or worse?

"using the gain pot" is using a program to process your signal. (we are talking itb mixing, right?)
Old 21st July 2011
  #10
Gear nut
 
SweetLossy's Avatar
 

Quote:
Originally Posted by LeibnizLover View Post
Doesn't using the gain pot also add a layer of dsp? i'm genuinely puzzled why one is better or worse?

"using the gain pot" is using a program to process your signal. (we are talking itb mixing, right?)
Normalization and Gain are the same DSP process: multiplication.
Normalization is a multiplication where we know the max limit: 0 dBFS.
Gain is a “free” factor multiplication, in 64 bit mixers you can go up to plus/minus 6000 dB.
So Normalization is a “fixed amount of gain” to a file. A fader represent a “variable amount of gain”.

The bad reputation of the Normalization process I think it comes from the 16 bit recording era, where recording the closest to 0 dBFS was mandatory to preserve the 96 dB dynamic range. For example, if you record at -18dBFS in 16 bit you have a 13 bit sound quality. If you apply Normalization to this file you will have noise problems.
Now with 24 bit files you can record safely at -18 dBFS, because you will have at least 126 dB of dynamic range. So it will be not a problem to normalize 24 bit files.

Also Normalization can be done virtually.
For example, Reaper has a Normalization option in its Item properties. This Normalization is done inside the mixer and it’s done virtually in 64 bit resolution. The sound does not suffer any change in resolution or noise. It’s completely transparent.

Quote:
Originally Posted by neilwilkes
Normalization adds an unnecessary layer of unwanted DSP to your audio file, with all the rounding errors & quantization distortions that implies.
Just say "no" to normalization & use the gain pot instead.
Doing it virtually, inside the DAW, you do not add errors. It’s a “linear” operation, so you can do it one thousand times and you will have the same sound.

If you are doing it physically, in the file, in 24 bit files the rounding errors are very, very low. In a -18dBFS peak file the errors will be at about -120 dBFS, far beyond the normal noise level of any recording.
It will be indistinguishable one multiplication more or two.

In 16 bit files things are different. It will depend a lot of its peak level. If your 16 bit file peaks at – 3dBFS, normalizing will not be a problem. If it’s a finalized file, it will be better to normalize it to -0.5 dBFS.

.
Old 21st July 2011
  #11
Gear nut
 
WhiteRhino's Avatar
 

Quote:
Originally Posted by SweetLossy View Post
Normalization and Gain are the same DSP process: multiplication.
Normalization is a multiplication where we know the max limit: 0 dBFS.
Gain is a “free” factor multiplication, in 64 bit mixers you can go up to plus/minus 6000 dB.
So Normalization is a “fixed amount of gain” to a file. A fader represent a “variable amount of gain”.

The bad reputation of the Normalization process I think it comes from the 16 bit recording era, where recording the closest to 0 dBFS was mandatory to preserve the 96 dB dynamic range. For example, if you record at -18dBFS in 16 bit you have a 13 bit sound quality. If you apply Normalization to this file you will have noise problems.
Now with 24 bit files you can record safely at -18 dBFS, because you will have at least 126 dB of dynamic range. So it will be not a problem to normalize 24 bit files.

Also Normalization can be done virtually.
For example, Reaper has a Normalization option in its Item properties. This Normalization is done inside the mixer and it’s done virtually in 64 bit resolution. The sound does not suffer any change in resolution or noise. It’s completely transparent.


Doing it virtually, inside the DAW, you do not add errors. It’s a “linear” operation, so you can do it one thousand times and you will have the same sound.

If you are doing it physically, in the file, in 24 bit files the rounding errors are very, very low. In a -18dBFS peak file the errors will be at about -120 dBFS, far beyond the normal noise level of any recording.
It will be indistinguishable one multiplication more or two.

In 16 bit files things are different. It will depend a lot of its peak level. If your 16 bit file peaks at – 3dBFS, normalizing will not be a problem. If it’s a finalized file, it will be better to normalize it to -0.5 dBFS.

.
Boom.
Old 21st July 2011
  #12
Lives for gear
So how is normalizing in wavelab with 24bit files? (i do it sometimes, not often)
Old 21st July 2011
  #13
Lives for gear
 

^^ just test it
Old 21st July 2011
  #14
pushing peaks to 0dbfs cant hurt further processing. when you put a plugin on that source it will probably be in the red.
Old 22nd July 2011
  #15
Lives for gear
 
Remeniz's Avatar
 

Quote:
Originally Posted by Adam Dempsey View Post
Points for brevity?
Normalization = gain (to a determined point).
Compression = gain reduction (then sometimes gain).

The former is done mostly needlessly, for level. The latter is done for the sound of it, ideally.
Compression will result in a gain-reduced, but compressed, signal but this obviously is down to what the compressor is doing: Reducing the difference between the peaks and quiet parts of audio. Then you'd use the output gain control to increase the level of the compressed signal.

Although i'd say that compression is used mostly for better control of an audio level, not just for 'flavor'.
Old 14th August 2011
  #16
Gear interested
 

Thread Starter
Great feedback (pun intended)

Hey, I just wanted to say thanks for all the great information. I especially appreciate your being concise and to the point.

I am now almost finished with my CD, a DIY, recorded, mixed, and mastered excursion in oddity. With this being my second self-produced LP, I feel like I have only begun to scratch the surface of audio engineering so do not be surprised when I come back with more technical questions.

In fact, I just thought of one.

I have a Macbook late 2008 and I was wondering if there is an option for temporarily disabling the CPU fan while tracking (in small bites of course) to bring the signal to noise ratio down to the bare minimum. My windows based computers each have bios options for running the CPU fan, but I am still a Mac newbie, enlighten me?
Old 14th August 2011
  #17
Lives for gear
 
Dr. Mordo's Avatar
 

I think normalization can actually be a very powerful tool.

One thing I have been doing is normalizing every track, then lowering the gain (on all files) to -6db. This way across the board I have a pretty consistent level going into my plugins. Initial gain staging is done. Takes 20 secs total.

Another trick that is pretty time consuming but can be very useful is to cut up a track into sections and normalize those. How many sections depends on what you are doing and how you want it to sound. For instance, you could cut a drum part into a million little pieces where every hit is separated. Then you normalize and reduce gain by 6db. Now every snare hit is the same volume with no compression necessary. Is this a good idea? Probly not. But it is one possible application. I have done this on bass drum parts, but not for every individual hit. I just balanced out different sections where the drummer wasn't kicking as hard.

Recently I had a vocal that was uneven because when the singer dipped into his lower register he got quieter. It wasn't intentional and had nothing to do with the dynamics of the song. So, I split it into sections and then normalized them. Then I dropped the gain on all by 6db and ran it all thru a channel strip and light compression. That way, instead of spending time programming faders or overcompressing the crap out of it, I just equalized the levels. Worked great.

I do think this thread should move to the newb forum, though.
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
 
Register FAQ Search Today's Posts Mark Forums Read

SEO by vBSEO ©2011, Crawlability, Inc.