Normalization Recommendation Needed!

[tommyzai]

I normalize all my recorded audio files in my DAW before mixing, sending out stems, etc. I've done this for over 30 years. I find normalizing helps keep some uniformity . . . it's part of my process.

I'm not talking about mastering . . . just concerned with the raw tracks on the DAW . . . in preparation for me editing/comping, then sending out track stems for someone else to mix.

In the past, I used -1.00 dBFS for everything, but after researching I wonder if there are better settings for type of normalization and target value.

Thanks for any constructive, relevant suggestions. I beg of you not to respond by arguing that normalization is not necessary and/or harmful. Thanks in advance.

[BertKoor]

I'm curious... what was your conclusion after doing some research?

[tommyzai]

I'm curious... what was your conclusion after doing some research?

No conclusion. That's why I created this thread. The information conflicted with one another. I became more confused. I think the problem is . . . some "experts" are talking about the loudness settings for various streaming services. All I'm after is what works best for normalizing the audio tracks within the DAW itself. Some say mixers like getting normalized stems, while others claim they hate it . . . not sure why they would hate having nice, neat, normalized levels to work with, but . . .

Anyway . . . here I am . . . asking . . . If normalizing tracks, what settings?

[No_Use]

Normalizing to -1 dB (peak) FS is not ideal for mixing imo, because if you sum a few tracks you'll certainly get over 0 dB and then have to turn it down somewhere.
Personally when getting stems for mixing (or start mixing my own crap) I normalize to say -14 LUFS (integrated or max. momentary, depends on the material) so I have nice enough headroom and the tracks have (more or less) similar perceived volume as starting point.

[BertKoor]

Have you asked the people on the actual receiving end?

For mixing purposes my preference would be the RMS level being roughly equal. But I'm an analog guy, I like my tracks hovering around 0dB VU which is about -18 dB fs.

Usually normalizing looks at the peak levels only. That makes pad sounds louder than percussive sounds.

Not what you want to hear: the levels will be changed in the mix anyway. Gain trim for the course level, fader for the real mixing balance.

Although futile, it's totally harmless.

[tommyzai]

I like having everything at optimal levels, then whoever mixes can mix with faders. My concern is . . . what type of normalization and was dB?

[BertKoor]

I think you're totally right. Set everything at what you think is the optimal level. Whoever mixes can mix with faders.

Conclusion: there is no problem here, so nothing to fix, so type & level don't really matter. Why can't you accept that?

[lfm]

Two things:
#1. Sometimes plugins don't like too hot level
- especially hardware emulating ones as I discovered

So have to use gain adjusting level hitting first plugin, or there is always overs shown on plugin.

So why first normalize, then having to lower gain anyway.

#2. If for some reason not needing more than +6 dBFs from fader, don't normalize
- faders usually has +6dB to increase level on a track

Today with at least 24 bit audio, and 32 bit float mixing in daw
- no need to worry about levels being -20 dB or lower even
- peaks at -12 dB is even recommended by many(as fader is at 0 dBFs

If needing more than +6 dB on fader, normalize.
- but daw usually have gain as well to adjust level hitting first plugin
- I would say, normalize never needed
- save the time for something else

Bonus thing:
Some even bother to split a clip in fractions removing silence etc.
- just makes selecting stuff more tedious
- like gating everything, use a gate plugin instead if noisy

[tommyzai]

When we record audio, no matter how careful we are, the levels will be inconsistent . . . track by track. Some will be -3 while others will be -5 and -1, etc. So, I like to bring all the raw audio peaks to a consistent level. This is what I'm/we're trying to do when recording in the first place, right? We try to record each track at optimal level. Besides this, it's nice to see track waveforms at about the same size. I realize dynamics will cause some performances to be low, then high, etc. But, at least they peak like the rest. Then, whoever mixes can work magic with faders. MANY normalize. I'm not doing something bizarre here.

So, my question is, when normalizing to raise audio peaks to optimal level, what type of normalization and what level? That's my question. I'm not asking if I should normalize, because I WILL and I know I am correct in doing so . . . . at least for me.

[BertKoor]

You have your answer already: there is no best way.
Normalising is just another stage of adjusting the volume

[tommyzai]

I'm deciding between Peak and True Peak.
I'm also deciding on -1dB, -2dB, -etc dB.

[gullfo]

consider this - "peak" tends to be a "very short time period" vs "true peak" which is "sample" level. so generally speaking my preference is limit -6db on true peak to avoid any possible clips. and generally normalize to -12 or -18 LUFS depending on the delivery specification on a per-track basis. the mix engineer can use their preferred gain and volume controls to get the mix they want. as noted in previous comments, simply normalizing based on peaks can skew the overall balance when the tracks are brought into a unity gain setup (e.g. pads being louder than other instruments) - and yes, they can turn it down - but it tends to be easier if the relative balances are close to their original intent.
i find tracks that are cranked up by simple normalization algorithms tend to sound harsh, and if i encounter those, i'll ask the recording engineer to re-send those tracks at the lower levels - wasting time when simply keep the levels lower will make things go faster (in my experience) than not. of course if the recording was already too hot and clipping distorting, then some extra work with RX or Acoustica to try to recover it.

[BertKoor]

Have you analysed the options? Any benefits or drawbacks to one or the other approach?

You researched, asked the community. No definitive answer. Perhaps because there is none?

Closest to a logical approach is not normalising, but rather gainstaging. This sets the targets to an average level, not so much looking at the peaks.

[tommyzai]

I'll leave gain-staging to the mixer. I just want to set up, clean up, and edit my tracks in preparation for sending out stems.

[BertKoor]

Ask your engineer then.