Who Uses Normalization and Why?

[Anderton]

There's a thread on the Cakewalk forums about normalization, and the general attitude seems to be that it's not a good thing. I totally disagree, I've found all kinds of uses for normalization and I'll get into them shortly...but I wanted to get other peoples' opinions.

[Mikeo]

It can bring an audio mix in to check, which can be nice when listening to a song at a lower volume, or even a very loud level.

 

There are as many reasons for normalizing tracks, as there are for not doing it. Can't fix a bad mix with it, but many think you can.

 

 

Dynamics are usually good and can be lost in a mix, but not always.

 

I look forward to your follow up article.

 

 

 

 

 

 

 

[MikeRivers]

W\hen I record a concert, I'm pretty conservative about the record level because I don't know what's coming. Often I'll end up with no peaks above -16 dBFS or so, so I'll normalize the whole file to peak at -3 dBFS (or maybe I'll just add 12 dB to the whole shebang) just to get it up to my more-or-less normal playback level. Often, that's all the processing I'll do other than edit out excessive dead air and inane nattering while someone's tuning or deciding what to play next.

 

There are two sensible arguments against normalization. One is that people tend to normalize to 0 dBFS and then try to compress and boost everything to match the peaks. With all the concern about "intersample peaks" exceening the 0 dBFS value these days (seems like mastering engineers and audiophiles always need something to worry about), normalizing to full scale opens the door to this problem even if you dont flatline the whole song.

 

Also, normalization boosts everything, including mic preamp noise. While this is probably quieter than turning up the playback volume on the analog side, it makes you realize that you should have been more attentive to gain structure when recording.

 

There's nothing wrong with an overall level boost if it doesn't do any harm.

[dboomer]

To your second point ... Yes it increases noise levels in absolute, but it doesn't change the signal to noise ratio of the single track. So you'll hear the same amount of noise once that track is blended into the mix exactly the same with either method.

 

When you are recording with 24bit depth you have 144dB of dynamic range, so I always record 12-20 dB down. But I always normalize so that I have room to work with thresholds of compressors, etc.

[AlamoJoe]

I think I could use a bit of normalization myself personally. I'll go away now and let this thread continue in it's serious vein.

 

[MikeRivers]

To your second point ... Yes it increases noise levels in absolute, but it doesn't change the signal to noise ratio of the single track. So you'll hear the same amount of noise once that track is blended into the mix exactly the same with either method.

 

I'm the "straight to stereo" guy here, so the track that's being normalized IS the mix.

 

Of course maybe the slick thing is to not normalize and let the listener turn up his own volume control to make it louder. Then we can blame HIS system for the increased noise level.

 

But then, nobody seem to know where the playback volume is any more.

 

[bookumdano4]

I normalize individual tracks in order to have easier gain-staging when I mix. I am a huge believer in normalizing. It doesn't degrade anything if you know what you're doing. I've never thought of normalizing as evil. Except brickwalling mixes. That I'm not into. But then, I'm not into compressors on the main bus either. Some stuff should be left for mastering.

 

Also..I would never buy a higher-gain pre to use an sm7b. Or for a low-level ribbon. I'd just use an sm7b with whatever pre or console I have .. and then normalize what I capture....although I'll never buy an sm7b.

[dboomer]

 

I'm the "straight to stereo" guy here, so the track that's being normalized IS the mix.

 

 

Again, no matter. Normalizing doesn't change the S/N of a recording. So unless your recording is way out of bounds you are going to incure some increase in absolute noise floor either in the track or the playback gear.

[philboking]

Normaling to 0 dB peak is fairly useless. It actually guarantees that at least one place will clip or be near clipping. Far more useful would be normalizing to -3 for mixes and -6 for tracks to be mixed. This allows working with them without having to weird things, such as setting a compressor threshold to -25 dB to get it to work.

 

Another option would be the ability to normalize the RMS level of a series of mixes to a common level, such as -16 or -18 dB, to simplify getting all the songs an an album to the same perceived loudness. This would probably require compressing peaks in the songs with high crest ratios (which is peak/RMS). At this point, we're getting into compression or limiting, and things are no longer simple...

[philboking]

I typically normalize a clip by using the normalize function immediately followed by turning down the selected item clip by 6dB so the is some headroom to EQ, compress,etc., without clipping.

[nat whilk II]

Normalizing simply considered as an overall volume boost does help overcome the fixed noise level inherent in playback.

 

Normalizing individual tracks makes mixing more a matter of moving sliders down rather than up. Again, some noise is reduced in the absolute ('tho not in the relative S/N of the single track, true.)

 

Normalizing can make working with compressors a lot easier - you can work in the compressor's sweet spot rather than out in the extremes of compressor settings.

 

Of course we should always rely on ears first, but on the odd occasion, I see something I wasn't really hearing that clearly. Normalizing makes the visual feedback of waveforms more informative at a glance. Especially if you have a complicated recording chain with effects - you can have a waveform that never reaches 0db but is nevertheless clipped or otherwise distorted from some bottleneck in the chain. You may not see that until you normalize.

 

Normalizing helps me realize just how maddening single-coil pickups can be in the studio

 

nat whilk ii

 

[UstadKhanAli]

If it's normalization specifically, I never use it because I can never think of a decent reason why I would do so. It never even occurs to me to use it.

 

If it's gain optimization, I use that all the time, not only on the Master Buss but occasionally on individual tracks (vocals, bass, etc.).

 

And if's like Alamo Joe, I'm happy not being "normalized".

[blue2blue]

I've certainly used normalization as a tool to facilitate working with a large number of tracks or recordings but now that 24 bit DAWs tend to have an in-channel 'trim' or 'gain' control, I use that to bring individual tracks into the nominally optimal range for working with plugin FX.

[WRGKMC]

All normalization does is resize the waveform amplitude based on peak information. If you have only one peak that is higher then the rest, it will normalize everything based on that one transient. Instead of getting an overall gain, you can have a decrease in volume.

 

Most recordings probably don't contain those large transient peaks but I have had them many times in the past, usually on bass and acoustic drums.

In those cases a clean limiter works better for gaining up the tracks.

 

I can say I don't use normalization as much as I used to. Mostly because I spent allot of time over the years targeting my ideal recording levels. Where I do get gain changes is when I tweak the tone of instruments being recorded or play in higher registers. A bass with allot of open strings will have louder waveforms then something played up the neck, or a deep bass preamp setting will be louder then something with more mids and less lows.

 

That's when I may normalize. I have it down to a fixed formula too.

I normalize driven guitars to -10~ -12 and clean guitar to -6~ -8.

Bass is always -5, and electric drums to -4.

 

Without allot of additional plugins this almost always gets me close to a -12~-14db RMS mix down level which is my ideal target for mastering.

 

It is also the ideal level for running any plugins I may choose. If my track levels are too low for example, something like a compressor wont gate in a realistic range. I want to be sure my threshold on a comp or limiter kicks in around -16db on a hot track and -24 on a weaker one. Any weaker and most plugins just don't work reliably as they should.

 

The other thing normalizing helps with is higher productivity. If I normalize guitars to -10, all the presets I built and saved at that level will be fairly close to matching that gain level. This can produce continuity between a series of recordings and save time having to build everything up from scratch.

 

The other place I use normalization is quick trimming noise. I set the level to infinity, select areas of the tracks where I want silence and normalize it to infinity. This can create noticeable jumps to silence so I usually include a second maneuver of creating a sharp volume envelope ramp into or out of a passage. I realize there are other methods of doing this but I do find it a quick way of silencing all tracks at the same time on intros and outs.

Since the length of an intro may vary an envelope alone can be hit or miss.

 

Volume automation works too but again, pulling down 16 channels on a controller to write those envelopes and writing those changes is neither quick or accurate. I can highlight all tracks with three mouse clicks highlight the areas to silence with a mouse sweep and bang I'm pretty much done except for tracks that need an envelope ramp. .

[E-money]

As easy as it is to automate levels nowadays, I've seldom had a need to normalize. I remember it being a big no-no back in the 90s, so I've avoided it whenever possible.