I am uploading videos to Youtube, and I check the loudness of every video to make it as close to 14lufs as possible. I use the loudness meter in Izotope RX7 for this. Yesterday I metered a file, set it to 14lufs. Today I noticed, that for another file that has nearly the same volume if you listen by headphone, RX7 reports 9lufs for that. How is that possible? No way that you need to reduce 4lufs because in that case the second file will be much much more silent than the one yesterday. What causes this oddity?

Here is the file metered at 14lufs:

Here is the file from today metered at 9lufs:

 

Because ... there are 2 values with different names. And that has a reason. Because you can not interchange volume (peaks I assume) and Loudness. Volume perceives and measures the peaks, the envelope so to say, loudness perceives and measures (roughly) the area between the envelope and the zero-line. That's very mathematically.

Imagine the following: You have a bass drum/kick at full bang. The volume reads 0db. You have another file that is full blown white noise. The volume reads 0db too with that file. Then you render them both to -14db LUFS. Listen then to both of them. Guess what? The kick drum file will be very much louder. Because LUFS measures an average, and in the case of the white noise file the average equals the peaks/envelope. So it is way more put down than the kick file. The reason is that in the kick file there is silence between the kicks, that lowers the average massively.

So the differences between your files come from the different "denseness" within the files. String pads (e.g.) have a higher denseness than a strummed guitar.

The other thing is: why do you chase the myth that your youtube uploads should measure -14db LUFS? Yes, I know, YouTube writes that. And every "knowledgeable" parrot does exactly this: paroting the YouTube blabber. YouTube takes care of that LUFS-thing. Make your renders/masters sound like you want it to sound. Do not care about YouTube or any other streaming service. They all take care of that. Maybe make a test (you did already, at least in parts): Render for different streaming platforms and listen to all this renders: they sound different! That is not what you want. Render/master one (!!) version as you like, don't care about what any of the "pros" says or parots. It's false.

What do you think goes through the head of Giles Martin, when he renders/masters the Beatles-albums at an average of -10db LUFS? He does that because he is an idiot and he doesn't know how to render for streaming services? Nope. Because he wants the songs to sound exactly sound like this. And if they are too loud ... iTunes, Spotify, YouTube, Deezer, whatever takes cares of the loudness. They do it anyway, regardless of you working soooo hard to meet the "target" levels.

Have a listen to Pink Floyds album "Wish You Were Here": Imagine what happens when you treat every single track on this album fitting the rules about LUFS? Try it ... (BTW: That is exactly the reason why there is an album replay gain and a track replay gain. Track replay gain applied to an album makes the album unlistenable, because it destroys the inner dynamics of the album.)

So ... think it over who you listen to and whose advice you follow and if it's even of interest what some say or advise ... goes for Nov. 5th also.

 

You got one file with a full frequency range but lot of dynamics, the other one is very compressed and no dynamic at all, but got a limited frequency range. So they seems to sound roughly the same volume because you hear more frequency on the dynamic one (Djo Poster) and the compressed (defao) is mostly mid range frequency, so it's perceive as loud as the clear one cause it's technically louder but got less frequency (so need more loudness to give the same energy, not sure if what I'm say is clear, I try to simplify).

Also, don't worry to much about Youtube LUFS, just master your track at the volume it feels good, and upload it as it is, just give a little headroom for peak (like -0,3) and it should be good. You can see how other track are "corrected" by Youtube (right clic "stats for nerds") and you'll see that a lot of video are way more loud than what is recommanded.

 

I think that's exactly the common misconception here. YouTube normalizes tracks that are louder to -14 LUFS, but YouTube never recommended to upload tracks with -14 LUFS. Don't "normalize" yourself. The normalization on platforms like YouTube, Spotify, etc. only changes the perceived loudness. It doesn't change the sound itself. But if you change a setting on your compressor, eq, or whatever, just to be closer to a specific LUFS value, you do change the sound – the dynamics, frequency response, balance... And why would you do this? You want your music to sound as good as possible.

 

You mentioned something I never understood: Youtube is asking for -14lufs but some videos are way more louder. Even me I uploaded videos that were 11lufs and they are still loud after I upload them to Youtube. Louder than the ones I measure 14 and upload.

 

YouTube is tweaking its audio settings and introducing the "Stable volume" feature.

New best YouTube feature (stable volume)

 

In two words, your perception.

First off, the first file isn't set to -1TP. After I've done that it has -13.2LUFS.
I pulled both files into my DAW and pulled the fader of the second one with the mouse and with my eyes closed down until I perceived them as equally loud.
Result, I pulled it down by 3.84dB and the remaining difference between these two files was then 0.3dB.

 

... is usually deactivated for music.

 

Okay, but if you say I should decide settings by ear, for youtube, what should be my reference point? So that I can compare all my future uploads for that? Can you suggest anything? Like a sample song or something?

 

Hello Sono, you will become more confident over time. Set your monitors to a level that you feel comfortable with, a sound level that doesn't hurt or annoy and doesn't damage yours. Please never change this Monitor setting unless you get old and your hearing gets worse.

Every mixer in the studio knows his monitors inside and out, only then will he be able to properly assess his mixing recording.

When you create a new sound, make a headroom of 8-12 dB, then mix your sound, add effects such as compressors etc. and stay strictly under 0 dB, ideally 1.2 dB under! Save your mixer chain. Every song is different, train your ear and over time you will get the hang of it.

 

Last edited: Oct 22, 2024 at 9:34 PM

Why anyone would adhere to -14LUFS is beyond me. Upload to whatever loudness it sounds good at and let the platform do the volume normalization.
You are going about it backwards. The loudness normalization is not for YOU. It's done FOR you.

Also, No Avenger seems to have answered your question regarding the two different files with different loudness, spectrum balance, composition, peak value, etc.

 

This! And only this. Forget about YT or other streaming platforms.
If you have a sparsly arranged ballad, -14LUFS can already be too high (not to speak of classical music), for a modern EDM, or Metal track -14 doesn't deliver the punch. You need to find the LUFS for each track where it sounds best to you.
Reference tracks OTOH can of course help.

 

Sorry, but I have too disagree with this statment.
Firstly, it's quite usefulful to play at very low level to check some balances secondly, you need to turn up the volume from time to time to get an impression of how it sounds this way. The perception of our ears depends on the loudness (in SPL), see https://sengpielaudio.com/Acoustics226-2003.pdf
Nonetheless, for most of the time you should listen to a comfy level for health reasons.

 

Oh, there's a misunderstanding, I meant the controls on the back of the monitoring monitors.

 

Ah, yes, that's correct. Usually they're not stepped or matched and you can easily ruin the L/R balance - if that's what you meant.
Levelwise it doesn't matter because you're usually pulling the level of your interface up and down all the time and without a SPL meter you won't find a specific (reference) level anyways.