To which value should I normalize my audio files?

[Alex Voellmer]

Could someone help me out on which level to set my audio to automate at?
I am getting confused by LUFS, dbFS, Peak, etc.....
I read about gain staging the other day and that most software emulations perform best when hit with a specific input volume. To my shame I have to admit that I had never really paid attantion to this.
To sum up:

1) which value should I set for normalizing audio prior to mixing?
2) Can anyone confirm (through testing) that normalized signals achieve a better sonic result when being sent into a software emulation of a hardware comp/eq/etc (compared to a non-normalized signal)?

Thx

[DMG68]

I don’t mean to not help you out, I’m sure others will reply. I don’t have time to get into it but you should google ‘gain staging’ and read more, it’s not going to be much different from what people are going to reply with here probably. You would generally use RMS for that, and most often around -18db set for 0 VU for analog emulated plugins.

[EdgarRothermich]

The internet is a blessing and a course. For beginners, there is so much information out there to learn for free, but it is all not curated. When you go to a school at least there is (hopefully) a screening process for the instructor and when working your ropes at a recording studio, at least, the engineers (hopefully) have skills, knowledge and experience.
Nowadays, every influencer-wannabe with a hunger for clicks can produce a video or comment on questions. Don't get me wrong, it is nice that people want to help and give advice, unfortunately, a lot of advice is questionable, or flat out wrong.
Many questions posted from people are just so basic that it is impossible to answer in one of two paragraphs. That's why you find all these quick cookie-cutter recipes. Do this, do that, use -6dB headroom, use -18dB headroom, use VU, use RMS, use PEAK,...
All that would be nice in a discussion to debunk them or at least talk about the pros and cons, or at least why. For a beginner, there is not much use for those recipes because their just blindly follow that without understanding why.

In the end, it is all about buzzwords (gain staging, normalizing, mastering).

Here is a just brief comment about that topic:
Gain Staging comes from the anal era where try to operate each component in its best level range. Now in the digital domain, you don' have to pay too much about the bottom and focus on the top (no clipping). Emulating Plugins might be an issue you have to pay attention to because the almighty 1500dB Dynamic Range of 32-bit float doesn't apply.
However, when your concern is measuring the level (especially of digital components), then VU is the worst tool you can use for (remember, VU stands for "Virtually Useless"). VU Meters are recommended so often on the Internet, almost romanticized because they look cool. If you use a proper Peak Meter (or True Peak Meter), then you don't need an artificial -18dB headroom. Remember what you want to measure. Level or Loudness. VU can't measure level correctly and for Loudness, they are a limited use because they have the wrong integration time and don't take the Equal Loudness Countour into account as proper Loudness Meter do. If you are an experienced sound engineer who grew up with VU Meter and knows how to interpret what they see, then that is perfect.

About the OP Question. If you are confused about LUFS, dBFS, Peak, and all that, then study more. If you are serious about engineering, then you can not be confused about those terms.
About 1) It depends. Just don't clip
About 2) It depends on the Plugin. If you don't clip it, then it should operate fine, unless it has special operation requirements. Read the instructions.

[Alex Voellmer]

This is a reply to DMG68. S.o. else postet while I was typing....

Thanks. That is what I read. I was just a little confused because my DAW (Studio One) has a default setting for normalizing audio at -12db. I understand that the incoming signal should be AROUND 0 db on a VU meter. And from what I remember that would correlate to -18db (but I am not sure if that was peak or rms or lufs).......... Didn´t find too many revealing articles about this topic through search engine. I probably was just not commited enough
Anyone who can sum his findings and insights up for me?

[thecontrolcentre]

i never normalise anything ... it usually means you end up having to turn it down again when mixing. Far better to record at a good level in the first place imho.

[Alex Voellmer]

BTW: I do know what LUFS, peak etc, are but I didn´t understand the advice that s.o. on youtube gave about this subject. His assumption was that VST effects react differently to the level of the incoming signal and hence produce different sonic results. The input knob usually found on a plugin compressor doesnt cure this. I had never heard of this and (as thecontrolcentre also mentioned) have been fine by just recording at proper levels.
I was just curious if anybody has come across this subject and could give advice on whether it is worth normalizing ALL tracks to a certain peak/rms/lufs level (and if so: which one?).

[vurt]

ive normalised in the past, if im mixing different source recordings.
ie sometimes i grab stuff on my phone, or from a friend, added to the daw.
it just makes sense in my head if using audio from different places to normalise them to a single place. (eve though, i know logically speaking, makes no difference (unless the levels are way off)

but in a single project, i tend not to bother (again unless i record something that for whatever reason doesn't provide enough signal...).

what problem are you trying to solve?
if there's no issue, there is no solution.

itb gain staging is only useful with certain analogue emulations, not something to worry about for every plugin.
the manual will inform if it's necessary, in fact most devs, will use the fact it's necessary as a selling point. so you will know if it's needed on a particular plug in

[EdgarRothermich]

This is a reply to DMG68. S.o. else postet while I was typing....

I understand that the incoming signal should be AROUND 0 db on a VU meter.

The first thing you should understand is NOT to use VU Meters for measuring Levels, especially in the digital domain. Use the proper tool like Sample Peak Meter or better True Peak Meter.
If you want to measure Loudness, then use a Loudness Meter, which is specifically designed for to emulate how the ear perceives loudness. VU Meter and RMS Meter are only close to how we perceive loudness because they show you the average level.

Forget about all the hype you read on the Internet about Gain Staging (especially if you blindly follow some advice without question why).
DAWs operate in 32-bit and 64-bit float, so the question regarding noise and distortion is irrelevant. The exception are those Plugins that either operate in fixed-point or have special requirements (read their instructions). The main purpose to get the levels on the individual Tracks in order is to have the Volume Fader at an optimum position for mixing that makes sense. If the Fader is at -40dB, then it is hard to make accurate level adjustments. Simple as that. It is a question about optimum workflow and not set your levels at some dB because the Internet tells you so.

[Tj Shredder]

As the question was about normalizing, only the final master needs to be normalized, meaning that it just doesn’t clip. If someone tells you that point is at -18 dB you know its a bullshitter who wants you to fail hard... If your peak is at 0 dB you risk that its going slightly above 0 if you do sample rate conversion in the digital domain or after the DAC, that usually won’t be a problem, but you are save to set it to something like -0.5 dB or 96%... The peak doesn’t tell anything at all about the loudness. Its your aesthetic decision how much dynamic you want. Normalizing does not change the dynamics, limiter/compressor does...

[Uriel Anthony]

Could someone help me out on which level to set my audio to automate at?
I am getting confused by LUFS, dbFS, Peak, etc.....
I read about gain staging the other day and that most software emulations perform best when hit with a specific input volume. To my shame I have to admit that I had never really paid attantion to this.
To sum up:

1) which value should I set for normalizing audio prior to mixing?
2) Can anyone confirm (through testing) that normalized signals achieve a better sonic result when being sent into a software emulation of a hardware comp/eq/etc (compared to a non-normalized signal)?

Thx

1) It has been answered in the 2nd post by DMG68 but let me rephrase it.
You need a VU meter (calibrated to 0VU = -18dBFS).
The needle should be dancing around 0VU most of the time during the loud passages. Occasical overshooting up to +3VU is fine.
But for transient heavy elements like drums and percs it should be lower around -10VU instead so the transients doesn't peak to 0dBFS (or whichever value you set for your VU meter's clipping LED to light up).
You should also be checking this between each plugin, not just at the beginning or end of the track chain.
Gain staging is simply to ensure proper sufficient headroom so it doesn't overload/clip the input of next plugin.

2) Yes. Especially more so for analog modelled plugins and even some (Ableton Live) stock plugins to a certain extend if you push it way beyond 0dBFS even though you are working in 32/64 bit floating point precision. The result (despite the fact that there's no clipping thanks to floating point calculations) is you will lose some low end and it will sound thinner and brittle as compared to when you proper gain stage. Why not try this experiment and do a blind A/B test yourself? Some say the analog looking VU meters are relics of the the past and has no place in the "digital domain". Let your own ears decide. If you can't hear it, then it doesn't matter whatever you do.

BTW, no you don't use automation to "normalize", that'd just kill all the dynamics in your tracks.

LUFS is only used for the final mastering stage and the person you are submitting your master to will give you the detailed specs that you should meet.

[EdgarRothermich]

1) It has been answered in the 2nd post by DMG68 but let me rephrase it.
You need a VU meter (calibrated to 0VU = -18dBFS).
The needle should be dancing around 0VU most of the time during the loud passages. Occasical overshooting up to +3VU is fine.

With all due respect, let me rephrase it. You should not give beginners advice that they need a VU meter (whatever calibration). And you don't need to be entertained by "dancing needles".

To properly measure the level of a signal (so it doesn't clip), use a proper meter like a Sample Peak Meter (or TurePeak Meter). To measure the loudness of a signal, use your ears or a Loudness Meter.
(VU Meters are 80 years old and invented for a different reason than measuring a digital signal in a DAW)

Loudness Meter is not just some esoteric thing that only a mastering engineer use. It also measure the level based on RMS (with three different integration times), but uses K-Filtering (nothing to do with the K-Meter!), so it is much more accurate to compare how you perceive loudness. VU Meters, on the other hand, give you a totally wrong value in the low range.

I don't want to repeat myself, but VU Meters are perfectly fine in the hands of experienced sound engineers, but they should not be recommended to inexperienced users just for sentimental reasons or the coolness factor of a dancing needle. Instead, users should be encouraged to learn and understand the underlying concepts of audio production and learn how to properly use the tools of the digital domain they are in.

[DMG68]

The things Edgar says are true. However, it doesn’t change anything regarding the analog emulated plugins you are asking about. They are modelled on old equipment, generally, that was designed for a signal around -18 db RMS. Yes, all outdated stuff. Do you have to use an RMS meter? No. You can use your ears. Feed your signal into your plugin at different levels and set it where it sounds best to you for each track. ‘Normalizing’ isn’t the proper word but I think you’re using the word you think best describes what you’re talking about, which is optimal level going in to analog emulated plugins.

In the end really, if you want to have a nice clean sound, you’d most likely feed analog emulated plugins much lower than you might think. I generally keep my peaks between -12 to -10 dbfs and it works out good for me, I rarely use any meters for mixing other than the peak meter. None of that changes what level is optimal for analog emulated plugins though, it’s just what I’ve found works for me. And fwiw I’m not interested in seeing you ‘fail hard’. If you want to use analog plugins and get more of an analog sound, then learn about that equipment and techniques used before digital and apply them in your daw. Do you need that to get a good sound? No. But you may need it to get the sound you want. You won’t go wrong to learn and use those techniques, and should definitely learn all the digital side too and what the differences are regarding various meters and what they measure. Everyone is different and what may be confusing for one inexperienced person may be the thing that clicks really good for another. Research it all, experiment a lot, and don’t take the word of any one of us alone.

[NAD]

If you're using Reaper with the SWS extension you can normalize your audio to a specified LUFS value which I use when I have a bunch of different files that I want to sound around the same loudness.

[Uriel Anthony]

1) It has been answered in the 2nd post by DMG68 but let me rephrase it.
You need a VU meter (calibrated to 0VU = -18dBFS).
The needle should be dancing around 0VU most of the time during the loud passages. Occasical overshooting up to +3VU is fine.

With all due respect, let me rephrase it. You should not give beginners advice that they need a VU meter (whatever calibration). And you don't need to be entertained by "dancing needles".

To properly measure the level of a signal (so it doesn't clip), use a proper meter like a Sample Peak Meter (or TurePeak Meter). To measure the loudness of a signal, use your ears or a Loudness Meter.
(VU Meters are 80 years old and invented for a different reason than measuring a digital signal in a DAW)

Loudness Meter is not just some esoteric thing that only a mastering engineer use. It also measure the level based on RMS (with three different integration times), but uses K-Filtering (nothing to do with the K-Meter!), so it is much more accurate to compare how you perceive loudness. VU Meters, on the other hand, give you a totally wrong value in the low range.

I don't want to repeat myself, but VU Meters are perfectly fine in the hands of experienced sound engineers, but they should not be recommended to inexperienced users just for sentimental reasons or the coolness factor of a dancing needle. Instead, users should be encouraged to learn and understand the underlying concepts of audio production and learn how to properly use the tools of the digital domain they are in.

You quoted me on my 1st answer but didn't touch on the 2nd one. To quote the OP's original 2nd question, he asked about signal levels being sent into a "software emulation of a hardware comp/eq/etc" to which a VU meter is recommended for that purpose rather than for some "sentimental reasons or the coolness factor of a dancing needle" (where do you even get such silly imagery from?!). FYI, I hate looking at the VU meter needle jumping all over the place but it gets the job done.

OP's 1st question was asking about calibration levels prior to mixing. He wasn't asking about measuring any form of loudness level, so in that case (ignoring his 2nd question), even any peak meter will suffice to ensure a certain digital headroom. So why are you recommending a whole bunch of loudness measuring plugins to further confuse him then? True-peaks isn't even a concern until the audio is leaving the DAW. OP is asking about mixing, not even mastering.

No, you don't need to repeat yourself. You're sounding like a hard sell salesman. The posts here are meant for answering the OP questions. Everyone's entitled to their own opinion, it's up to the OP to decide which piece of info he wants to absorb or try out but you seem to be picking on everyone every time the term "VU" is mentioned by quoting them and insisting that your answer is definitely the right one.

With due respect, I have no interest in anyone who poses as an audio guru on YT, that is until you decided to quote me. So please, unless you have anything new and constructive to contribute to answering the OP's questions I suggest you keep your biases to yourself instead of continuing to trash on other user's posts and hijacking the OP's thread or bombarding the OP with a whole bunch of info that he wasn't even asking for further confusing him. Talking about K-filtering to the OP, a beginner? Seriously? Nobody's even talking about "value in the low range" here.

It seems that you don't even understand that VU meters are for observing how hot a signal is (on the long term, or aka average) relative to the circuitry of an analog (or analog modelled) audio processor which is directly related to OP's 2nd question. VU meters were never meant for loudness measurements. In the old days they will use RMS or PPM meters for that instead.

users should be encouraged to learn and understand the underlying concepts of audio production and learn how to properly use the tools of the digital domain they are in.

I think you're falling on your own point here already by denying the OP a chance to learn about VU meters plugins, or are you going to go on and argue that VU meter plugins are not a "tool of the digital domain"?


I've made all my points and have nothing else to contribute to the OP's answer so I will not post in this thread any further.

[Googly Smythe]

https://www.learndigitalaudio.com/normalize-audio
"normalisation" isn't "gain staging".