Should I normalise audio after recording?

[dickiefunk]

Hi. I use Cubase SX3.1 for my DAW. Should I normalise the audio track/tracks after recording or should I leave this till after I've exported the finished song?

What are the advantages and disadvantages of doing this?

Thanks

[Farview]

There is no reason to normalize individual tracks...EVER!

You will just end up having to turn the channel faders down to compensate when you mix.

Just record the tracks at the right level and mix the song.

[dickiefunk]

Yeah this is what I figured. Wouldn't normalising tracks risk adding noise?

[Robert D]

It's deja vu all over again......and again..........and again......

This topic has been discussed extensively, so use the search button and all will be revealed.

[LeeRosario]

Originally Posted by dickiefunk

Yeah this is what I figured. Wouldn't normalising tracks risk adding noise?

well what you're doing is bringing out whatever digital noise you'd have near the bottom of the noise floor.

For example, lets say you record some vocals with about 3 db of headroom and you decide to normalize all your tracks to have 6db of headroom. You've permanently lost a very small part of your signal to that digital noisefloor and you can never bring that back. The difference is subtle, but its there.

To renormalize that to 3db of headroom would then sound slightly different.

It dosn't translate so much like analog hiss, but more like this "not as smooth as the original" digital grundge. Once your signal disapears into your noisefloor, you can't really bring it back out of nothing.

I guess a good way to test the theory is by taking a perfectly healthy signal, normalizing it to something extreme like -34db and then renormalize it back and then hear what it sounds like.

I'm only going off what I've heard in the past.

[XLR]

Originally Posted by Farview

There is no reason to normalize individual tracks...EVER!...

+1
.

[IronFlippy]

Normalizing is quantizing the signal again, which brings up the noise floor, in addition to it being brought up by the actual normalize. Normalizing is an absolute last tool to use if you have an extremely weak signal and you can't go back and rerecord it. Even then, I see no reason to quantize to 0db.

Listen to a fade out on a cd. Loop the very end of it and crank your monitors. Assuming no noise from the monitors themselves, you will hear a nasty hiss. That's the noise floor. Don't let that be noticeable.

[7string]

I think if your levels are too low AFTER you've recorded something then you should look at your recording process. A properly recorded track should not have to be normalized.

After you've exported the finished song you should use compression or a good limiter to bring the levels up.

[TerraMortim]

Originally Posted by Farview

There is no reason to normalize individual tracks...EVER!

You will just end up having to turn the channel faders down to compensate when you mix.

Just record the tracks at the right level and mix the song.

not entirely true. I have found that when working with virtual guitar amps, it can effect the timbre of the virtual amp after normalizing (similar in way to how the volume of the guitar going into an amp would effect the timbre of the amp... And there are a few times where I want to slam the shit out of a part with something like PSP vintage warmer, which having it normalized can bring a different sound to if you use strictly the controls in the plugin...but yeah, generally, it won't do that much for you.

[dickiefunk]

Thanks for the replies guys. What I've been doing in the past is I've been normalising all my audio tracks because a friend told me that's what I needed to do! I wasn't pleased with how it was sounding so I thought it's about time I found out some more!
I've only been using EQ on the master bus. After I exported the finished songs, I have been normalising them to 0dB on Wavosaur to make sure they're playing at the same volume. Is this a good idea?

[SouthSIDE Glen]

Originally Posted by dickiefunk

After I exported the finished songs, I have been normalising them to 0dB on Wavosaur to make sure they're playing at the same volume. Is this a good idea?

While that probably won't hurt anything, normalizing the peaks like that does nothing to guarantee that two songs will be heard at the same volume.

Perceived volume is based far more on the density and the average volume of of a song than it is by the level of the peaks. It's possible to have two songs that are both peak normalized to 0dB and yet sound several dB different in overall volume.

G.

[Farview]

Originally Posted by TerraMortim

not entirely true. I have found that when working with virtual guitar amps, it can effect the timbre of the virtual amp after normalizing (similar in way to how the volume of the guitar going into an amp would effect the timbre of the amp... And there are a few times where I want to slam the shit out of a part with something like PSP vintage warmer, which having it normalized can bring a different sound to if you use strictly the controls in the plugin...but yeah, generally, it won't do that much for you.

Your mixer doesn't have trim control? I would think it would be better to raise the volume in 32 bit float than to normalize in 24 bit.

I never normalize each individual track, what's the point of that?

I do normalize on mix down a lot though either because I don't want to use a limiter or I'm just lazy.

[TerraMortim]

Originally Posted by Farview

Your mixer doesn't have trim control? I would think it would be better to raise the volume in 32 bit float than to normalize in 24 bit.

yeah, it would be. I don't have a mixer tho, just mix totally in the box, if I throw the fader upwards it doesn't do anything to effect the sound going into the plugins, simply just raising the volume of them, because the effects are all pre-fader. I could send it to an aux in a send but that wouldn't make so much sense for a guitar amp plugin. Would the input level control be a bit noisier than anything in the channel? I suppose throwing a compressor with gain control on it to raise the level would work too, but wouldn't it be easier just the normalize, rather than use another plugin to zap resources?

Good point about the 32bit float vs. 24 bit tho..didn't think about that. I'm wondering though, how noticeable the difference would be, once you get everything mixed together with effects and all that. I suppose maybe you'd notice if you normalize every track, but I usually just will do one or two tracks here and there if the compressor doesn't quite saturate enough, or if I want a similar saturation/over distortion effect with the guitar amp plugin.

[Farview]

I'm talking about in the box. There is probably a trim control on the channel strip in your DAW.


All the talk about noise floor (as long as you aren't having the normalizer add 60db of gain) and the 24 bit vs. 32 bit float are just little details that can add up. On any single track, it really won't matter. But, when it becomes common practice and you start doing it on a bunch of tracks ('cause that's what you do), that's when you run into trouble.

[NL5]

Nevermind....

[NYMorningstar]

Originally Posted by dickiefunk

Hi. I use Cubase SX3.1 for my DAW. Should I normalise the audio track/tracks after recording or should I leave this till after I've exported the finished song?

What are the advantages and disadvantages of doing this?

Thanks

One of the disadvantages of any real time processing once you've exported the tracks to 16 bit is that you need to apply dither again. This means you are adding noise to your mix twice.

Ipso facoto, you should not change the output level after dithering the signal.