Tell me about normalization - what is it and what does it do?

  • Thread starter Thread starter ido1957
  • Start date Start date
It brings everything up, and sets the loudest peak at 0 dBFS. It can make things somewhat louder, but it won't even out all the tracks.
 
There's a few different types of normalizers.

There are level normalizers in the multi-pass category that have a tool that will analyze your wave file, find the highest peak and raise it to what you tell it, e.g., 0 dbFS like Harvey suggested. These require two or more passes.


There are loudness normalizers in the multi-pass category that allow you to specify the loudness(RMS) of a file. You have to be careful with those because if set too high you may cause clipping. It's a good idea to have a peak limiter(the Peak Master plugin) as part of that process. Again, a number of passes are needed

There are pan normalizers that will ensure that both channels of a stereo file have the same level or loudness.

There are normalizers in the Ultra-pass category like the Meta Normalizer that allows you to give all the files in a batch the same level after processing. It requires two passes, the first to analyze all the files and the second to process them.

Some normalizers have an 'only if clipping option' too which will lower the levels when appropriate.
 
It was a semi-handy thing to have 15 years ago.

Eh, come to think of it, no it wasn't.

Haha. ;) As it relates to 99% of music recordings, I agree with John here.

However, there are many instances where normalization of the audio file still applies. In voiceover work, nearly every client or post facility requires that a raw voice track is normalized to a "standard" volume level. A couple of years ago, that meant -3dB; nowadays it's often -1dB (which is ridiculous as it often degrades the quality of the audio). Personal feelings aside, most clients won't consider or accept anything that isn't normalized to at least -3dB. As NYM points out, I - and most others who do commercial VO - use a multi-pass normalizer that first scans the audio file to determine the loudest peaks, then actually processes the audio data on the second pass. The whole process takes less than 20 seconds or so for a standard 30-second radio read.
 
In concept, it "can" be used to level the sound of a CD so that everything will sound approximately the same volume. In my very limited experience, it just makes the CD sound like hell because it usually butchers the peaks and yields a bunch of square waves. Not a good tool for that use. In fact, I'm not sure what it is good for....

A better approach to leveling might be to tailor each tune on a CD to a known standard (-3dB or whatever) individually and then listen to the work as a whole. When it sounds good, stop messing with it. :D
 
In concept, it "can" be used to level the sound of a CD so that everything will sound approximately the same volume. In my very limited experience, it just makes the CD sound like hell because it usually butchers the peaks and yields a bunch of square waves. Not a good tool for that use. In fact, I'm not sure what it is good for....

This isn't what normalizing does. You might have it switched up with something else.
 
Normalizing is a good way to make sure your mixes hit those brickwall limiters super hard because it don't sound good unless you are louder than the other guy. :D
 
I've used two different "versions" of normalizers. One I think was in Audacity or Goldwave, and the other in Adobe Audition. The first kind was basically a compressor/limiter. It would smash the dynamics and rasie the overall volume to whatever you specify.

The other one that I use all the time is just a quick way to bring your highest peak to a specified volume level. Once you do that you can tell how big the gap is between the average volume level and the loudest peak. Then you can adjust it to get the overall wav closer to where it should be. I don;t see any reason why it wouldn't be useful now. I use it all the time.
 
Haha. ;) As it relates to 99% of music recordings, I agree with John here.

However, there are many instances where normalization of the audio file still applies. In voiceover work, nearly every client or post facility requires that a raw voice track is normalized to a "standard" volume level. A couple of years ago, that meant -3dB; nowadays it's often -1dB (which is ridiculous as it often degrades the quality of the audio). Personal feelings aside, most clients won't consider or accept anything that isn't normalized to at least -3dB. As NYM points out, I - and most others who do commercial VO - use a multi-pass normalizer that first scans the audio file to determine the loudest peaks, then actually processes the audio data on the second pass. The whole process takes less than 20 seconds or so for a standard 30-second radio read.

not at my studio (which is in your backyard ;) )...I turn all "normalized" audio way down anyway. It's really pointless to send it normalized as I prefer to work with the raw recordings. I'm more interested in the RMS levels and how my levels will translate in the analog world (ie. when bumped to Beta SP)
 
:rolleyes: I have this kind of "related" question about normalizing.


WHY is the audio so BAD on television? Each time I hear a different channel, I get a different sound. Some loud. Some fine. Some too soft. Needless to say, some of the actors cannot articulate very well which is just garbled.

I wonder if the change to digital machinations will turn out "better?" The signals heard on a TV set will still be Analog sounds but better enhanced via digital signals?

I do notice that most tv commercials are "crystal clear" most of the time. Thus, I don't think my ears are losing my mind yet.

Any thoughts?

Green Hornet:cool:
 
not at my studio (which is in your backyard ;) )...I turn all "normalized" audio way down anyway. It's really pointless to send it normalized as I prefer to work with the raw recordings. I'm more interested in the RMS levels and how my levels will translate in the analog world (ie. when bumped to Beta SP)

Benny,
Odd, as this has been the norm for me for quite some time. Everybody seems to want the final product @ -3dB... I could see if I were contributing only a piece of someone else's production that they might want to do the leveling later, but I've yet to run across this in practice.

BTW, where are you in KC? I'm at roughly 8th & Central...
 
:I do notice that most tv commercials are "crystal clear" most of the time. Thus, I don't think my ears are losing my mind yet.

Any thoughts?

Green Hornet:cool:
They certainly seem to turn up the volume when broadcasting commercials. I wonder if they've done any studies on the success of TV commercials relating to loudness:D:D

Until they come up with an industry standard I'm afraid we will continue to experience this disparity in volume.
 
:rolleyes: I have this kind of "related" question about normalizing.

WHY is the audio so BAD on television? Each time I hear a different channel, I get a different sound. Some loud. Some fine. Some too soft. Needless to say, some of the actors cannot articulate very well which is just garbled.

I wonder if the change to digital machinations will turn out "better?" The signals heard on a TV set will still be Analog sounds but better enhanced via digital signals?

I do notice that most tv commercials are "crystal clear" most of the time. Thus, I don't think my ears are losing my mind yet.

Any thoughts?

Green Hornet:cool:


it's because the TV shows have dynamics...while the commercials don't. If you're going from a show like 24 where the actors go from whispering in one scene to blowing up cars in another, the engineers want to keep that dynamic range in there. It's what gives drama. Then when you pop into the commercials you have it riding at a single dynamic...loud. Stations actually have peak limiters...but of course those peak limiters aren't catching the average levels. So it's the contrast in perceived volume that's annoying you.

Many stations do have a set standard and require to receive that standard from studios.
Dolby Labs has introduced Dialnorm for this exact problem.

Which brings me back to what KineticSound was talking about, no studio would accept -3dB - -1dB peaks from me. They would either reject them or have to turn them down on their end. I actually find my mixes sound loudest on TV when I DON'T try and compress them a push them up to full scale. My guess is this is because the engineers on the station end don't have to turn my mixes down and also my mixes aren't hitting the broadcast limiters. Funny that the audio for video world has more of an accepted standard than the music world.




BTW, where are you in KC? I'm at roughly 8th & Central...

Our studio is south from you then...in the Waldo area
I live over on the KS side, though :)
Are you in the lofts down there?


giving clients mixes (whether strictly VO or mixed) based off of a standard peak levels I find is pointless for me. Mostly because of the above mentioned issue NYMorningStar talked about. If I happen to have even one single peak (a mouth pop, drum hit, etc.) hit at -3.1dB...and I try and normalize to -3dB, my whole mix is moving up +.1dB. If I had the peak at -1dB and normalized to -3dB...my mix would turn DOWN. Normalizing is just an automated volume movement based upon the highest peak. It doesn't do anything that my turning up the fader wouldn't do. Peak levels aren't nearly as important to follow than RMS levels when you're trying to measure perceived loudness.
 
Last edited:
Back
Top