Utterly confused re metering and levels

[ChamomileShark]

I'm using Cubase 7.5 and after reading about levels I've become totally confused.

I've had a look at the Cubase 7 manual and I've had to guess at what the two different legends are - I think the one within the meter is the peak level which has 0 at it's highest mark and RMS is at the side which goes up to +6. Have I got that right?

I've read that I should aim for -18dbs when recording - is that peak level or RMS?

Finally, what post fader level should I be looking for when I convert to .wav? And is that peak level or RMS?

thanks!

[DJ Warmonger]

I've read that I should aim for -18dbs when recording - is that peak level or RMS?

It only makes sense to limit peak level if our track will be passed for mastering. Still, -18 dB sounds absurd and you'll be fine with -3 as well. Just make sure the soudn doesn't clip. Still, different engineers have different preferences.

Don't know what "dBs" means in a context of sound and neither does Wiki. Did you mean "dBfs" - relative to full scale?

[JCJR]

In the old days shortly after the invention of fire, people used mixers equipped with slow mechanical needle meters whose ballistics were similar to average or rms readings. Both simple arithmetic averaging or the slightly more involved rms method look for an average amplitude or power. For instance, if a sine wave peaks at +1 volt at its highest point, the rms method of averaging would measure 0.707 volts.

To further complicate matters in early paleolithic recording studios, the tape recorder meters were calibrated so that a certain acceptable distortion level would read 0 dB on the primitive recorder level meters, but the tape recorder didn't sound too awful on some instruments if the peaks would occasionally swing to +3 or even +6 or higher.

So then comes digital, which sounds gawd awful distorted if you exceed the max numeric limit even on brief peaks (in my opinion). Even before digital recorders, came along fast LED or electroluminescent meters which could meter peaks, in addition to various averaged levels. The first digital recorder meters were calibrated with 0 dB at the top. That was the brick wall top, and going beyond that level yields instant mega-suck.

So when early cavemen connected old mixers with mechanical meters to digital recorders, the cavemen were long accustomed to letting the meters occasionally bounce way into the red.

This problem was typically solved by running a sine wave into the mixer, adjusting the mixer level to 0 dB on the paleolithic meters, and then adjust the fast meter of the digital recorder to -18 dB. That guaranteed that even if Alley Oop records cave rock solid in the red (slow average metering, with transients lots higher than the +6 dB average which the meters might indicate), then it was unlikely that the levels hitting the digital recorder would exceed the 0 dB mega-suck level on the recorder. So Alley Oop or Ogg could mix confidently while watching the mixer meters, just like with an old studor or mci tape machine, without having to constantly look over his shoulder to watch the fast meters on the digital recorder, to make sure the recorder ain't clipping.

The advantage of the slow metering is that it seems to sometimes better correlate with how the ear perceives sound. A mastering engineer might belabor the fine points.

I just always use fast meters, keep an eye on the peaks, and record as hot as it will get without ever lighting the clip indicator.

[thecontrolcentre]

+1

[Compyfox]

I'm using Cubase 7.5 and after reading about levels I've become totally confused.

There is no need to be confused - unless you read the manual which still wrongly writes "VU" as description for the peak meter.

I've had a look at the Cubase 7 manual and I've had to guess at what the two different legends are - I think the one within the meter is the peak level which has 0 at it's highest mark and RMS is at the side which goes up to +6. Have I got that right?

Wait - it depends on WHAT meter you chose.

If it's the Digital one, then the bargraph in the middle is showing VU/RMS in dBFS (300ms integration time, unweighted), the outer meters show digital peak.

If you selected the K-System meters in the Control Room, the horizontal black bars hovering inside the colored bars are RMS (at 600ms integration time, unweighted), everything else is digital max. Again, dBFS.

The other bargraphs in the Control Room are DIN, NORDIC and BRITISH, which are digital meters but with a ballistic of 5ms to 10ms and a custom reference level.

The bargraphs in the mix console itself (per channel) are digital (dBFS).

I've read that I should aim for -18dbs when recording - is that peak level or RMS?

RMS (or correctly said "average signal strength", or AVG in short - RMS meters are wrongly tagged RMS as realtime meters).

Take a dive into my KVR marks, I wrote a couple of posts about this.

The KVR Mix Challenge Rules FAQ also links to some of my posts.

Finally, what post fader level should I be looking for when I convert to .wav? And is that peak level or RMS?

Everyone does different, but I work at a reference level of -18dBFS avg (RMS realtime) / 0 VU (reference at -18dBFS) and a maximum digital peak at -3dBFS on he summing bus.


I also use this reference level (-18dBFS) while leveling in individual signals (per channels strip), but won't let the digital max peak exceed -6dBFS. Not only can I integrate hardware better, "emulation type" plugins also work at optimum levels.


BTW:
-18dBFS as reference level is common today - but after all these years I still haven't found out who started to mention/recommend this particular level. I also wrote my thought on this in one of my many posts on KVR. (again, see the FAQ's of the KVR Mix Challenge guidelines)

But good post by JCJR nonetheless.

[JCJR]

TW:
-18dBFS as reference level is common today - but after all these years I still haven't found out who started to mention/recommend this particular level.

I haven't a clue, but do recall that it was already a common recommendation shortly after sony pcm adapters + beta video machines became the first mainstream digital recorders for common studio use. Of course, before that was the time before pcm adapters/beta, and post- mass market cd releases, when mastering engineers had to have a much more expensive pro video deck and equipment to make masters destined for the cd manufacturers. Perhaps that is the era that the -18 dB idea first originated, dunno.

I only recall that time bracket because I was a fairly early adopter of sony pcm. Pre-internet on compuserve audio forums, and the PAN network, glorified bbs systems, the well-heeled famous mastering engineers were dispensing such advice.

[Compyfox]

The fun thing is...
I measured dBFS values after sending in a +4dBu studio level (1,227V) sine wave (1k) into various ADC's. Depending on the ADC, this can vary between -15dBFS and -12dBFS on the digital meter. My current ADC's "ideal worklevel" (+4dBU) is somewhere around -15,3dBFS (at least it was last time I measured this - which was 2-3 years ago) if I'm really pedantic about this. But I'm using -18dBFS anyway. So I'm actually slightly "under-loading" my gear. Which I even prefer these days - more headroom, with slight tradeoffs like noise if I implement really old digital outboard FX modules.


Sengpiel Audio summed it up pretty well:
"Notice - Comparing dBu and dBFS: There is really no fixed world standard like e.g. −20 dBFS = +4 dBu = 0dBVU.
The digital peak scale is not equivalent to the analog RMS scale.

You can never match dBFS and dBu"

So the reference levels by EBU (-18dBFS) and STMPE (-20dBFS) seem to have different roots.
But nobody can clearly explain "why" it's these values. And trust me, I'm on to this for over 6 years now!


Rule of thumb - always measure twice, never trust anything that is given to you in the product description or even the manual. Unless we talk about certain developers that triple measure their gear before they release it.

[Mister Natural]

My simple method these days is to not let any step of the chain - whether instrument input, vst efx, channel send, verb returns, etc - nothing peaks beyond -6db

[lfm]

Maybe this article is useful:
http://www.soundonsound.com/sos/may00/a ... igital.htm

[JCJR]

The fun thing is...
I measured dBFS values after sending in a +4dBu studio level (1,227V) sine wave (1k) into various ADC's. Depending on the ADC, this can vary between -15dBFS and -12dBFS on the digital meter. My current ADC's "ideal worklevel" (+4dBU) is somewhere around -15,3dBFS (at least it was last time I measured this - which was 2-3 years ago) if I'm really pedantic about this. But I'm using -18dBFS anyway. So I'm actually slightly "under-loading" my gear. Which I even prefer these days - more headroom, with slight tradeoffs like noise if I implement really old digital outboard FX modules.
[...]
So the reference levels by EBU (-18dBFS) and STMPE (-20dBFS) seem to have different roots.
But nobody can clearly explain "why" it's these values. And trust me, I'm on to this for over 6 years now!

Rule of thumb - always measure twice, never trust anything that is given to you in the product description or even the manual. Unless we talk about certain developers that triple measure their gear before they release it.

Thanks Compyfox, very interesting. I'm ignorant of new studio practices. Haven't tracked much for at least a decade but might get motivated to pick it back up for fun. Lately been deriving sufficient enjoyment "getting ready to work" but if I never actually do any work then that's ok as well.

Actually, am most likely equally ignorant of the fine points of old studio practices. The early editions of "Modern Recording Techniques" devoted a fair number of pages on minutia of studio gear calibration. I've done my share of calibrations over the years, but some folks did that and little else for decades. Practice makes perfect.

The author seemed to imply that a recording engineer should know how to routinely do that himself, but dunno how disciplined was the average mid level practitioner in the field.

Some of the better people at recording good sounding music happened to be fairly non technical, so perhaps most calibration was done by occasional visits by a technician. The big recording centers, nashville, la, ny city had techs who made a living at that. Perhaps analogous to routine visits by the piano tuner. Some facilities regularly having the piano tuned, and others only calling a tuner when the piano has become unplayable.

Am too ignorant to know the "big centers" outside usa. In the boondock usa studios, would guess that if the studio owner or engineer didn't do calibration, then it rarely got done.

The studio equipment usually came with detailed service manuals and calibration points thruout. For instance one might connect a sine wave to each channel, adjust the sine input level using a (hopefully precise) ac voltmeter, then adjust the gross gain of the channel and the meter to read true. And make sure the meter on the recorder exactly matches the meter on the board, etc. Plus level, pre- and de- emphasis, bias and distortion calibrations on the recorder (using the brand and batch of tape which one is currently using), etc.

A person could get very obsessive if so motivated. Though there was a precision limit, considering that much of the old gear was designed using slide rules only good for 3 or 4 significant figures. Which often happened to be "close enough" in practice.

Am just saying, in the old days if a fella wanted a device to be "exactly" at a certain level, the equipment could be adjusted accordingly. Though it was horribly expensive equipment. At worst case, if a level was off and a person was motivated enough, he might calculate and construct precision pads, using hand-selected resistors, to fix it. Though even with 1 percent hand selected metal film resistors, the value could change (unpredictably) from the heat of soldering the circuit, so absolute precision was difficult.

Am sure if you are doing it for money you are wise to conform to whatever standards the industry demands nowadays. For interoperability if nothing else.

Perhaps I should re-examine my habit of merely running as hot as practical while remaining clip-free. But for instance- Legato electric bass has a low crest factor and an rms level can be rather high without clipping, wheras such as tamborine or cowbell has a huge crest factor and in some cases might clip even at rms meter levels LOWER than -18 dB.

In analog there was huge noise penalty for recording too quiet. Even in 16 bit, a legato, low crest factor melody recorded at -18 dB is basically no better than a ~ 13 bit sample. So IMO it was beneficial to go "as loud as peaks allow" even on 16 bit digital recorders.

Even with 24 bits, dunno the actual practical bit depth of modern esoteric expensive converters, but if a converter is only "linear" down to 19 or 20 bits, giving away up to 3 bits on the top (-18 dB on a low crest factor track) sacrifices up to 3 bits on the bottom.

Additionally, there is ordinary analog thermal noise circuitry surrounding even the finest converter. To say nothing of the noise in the chain before and after the converter. Short of operating the analog electronics in a liquid helium bath, the thermal noise is difficult to reduce beyond a certain (rather high) floor.

And a matter of convenience-- Though I haven't done much recent recording, when obsessively polishing my own silly songs I do a LOT of digital wave editing. Perhaps some softwares are better/more convenient, but a quiet track in many digital audio editor windows, is a tiny little squiggle occupying a big mostly empty view panel. I find that terribly inconvenient for extensive surgery on a track, doing dozens or even hundreds of edits on a track, attempting to turn the sows ear into a silk purse.

Tracks recorded "as loud as peaks allow" are much easier to micro edit. For instance, if wanting to precisely remove a very quiet accident between notes, the accident might be a tiny feature even in a loud track. In a quiet track where the loudest notes are small squiggles, the quiet accidents can be invisible without taking the time to zoom in.

Anyway, am not arguing against standards. Merely explaining my rationale.

[codec_spurt]

I've read that I should aim for -18dbs when recording - is that peak level or RMS?

Finally, what post fader level should I be looking for when I convert to .wav? And is that peak level or RMS?

thanks!

You want to be looking at both peak and average. Both tell a different story. It is possible to have very low peaks but the sound is through the roof with the average. It is possible to have very high overloading peaks, but the general average signal is still very quiet. You want a balance of these two, hence the need for metering on both the peak and average.


Listen to CompyFox. He knows what he's talking about.

Me? I don't mind pulling the master fader down when I forgot to gain stage all the other faders. Same result. -3dB.

But where I can, and more often that not, I have everything pulled down to at least -12dB or -18dB. NO loss. When recording at 24 bit, and dont' forget your daw is working internally these days at either 32 bit or 64 bit even. They don't need to, but marketing dictates. This means you can have the fader way way low or way way high. Just make sure the plugin itself is not 'clipping' and distorting. These are two distinct things. You only need to be within range with this. Not too high, not too low. I call it the Porridge zone.

No need to pull the master fader down then and hack it.


But still, there are some very quiet and some very loud VSTs on the market and you will need to 'trim' it. Up or down. Season to taste. Most behave well, but if you use a lot of them, then some will be unruly. Allow for this.

But that is a good ball park. I have VUMT by Klanghelm and this helps in not only being able to see RMS and peak levels, but you can use it to trim the signal. It lets you know when a signal has gone way too far, but even then, this can be allowed for the odd peak.

Bass that might only be at a lower level average will take up more of your signal bandwitdth than the odd peak of the snare crack. It is disproportionate, so you can not say this rule goes equal across the board for all frequencies. You need to learn this. It is one of the most important things you will ever learn. Take a week out of your life, at least.

For the odd snare crack, limit it with a clipper and not a limiter. http://www.gvst.co.uk/gclip.htm is a god send. You can use this on tracks easily. You can use on master too but it takes a bit more knowledge. Not difficult though.

We are all assuming you are talking about digital recording because for analogue those figures would be very very different. More information please about your set up and what you are trying to achieve.

[codec_spurt]

I wrote a big bunch of notes for this and I can send them to you if you want. But I'm not sure they would help or be any better than you figuring it out for yourself.

You need these tools:

Klanghelm VUMT = costs like the price of a Good beer. (In my country at least)
ToneBoosters EBULoudness = http://www.toneboosters.com/tb-ebuloudness/ - Free for our purposes.

You don't need to buy EBULoudness unless you want to coz it will work fine for your purposes. It just won't save settings for later (which is useless generally for this type of thing). Essentially it is free.

Also see Sonalksis FreeG https://www.sonalksis.com/freeg.htm

They make it a nightmare to download and they have the worst installation software that will abuse your system. Be warned. Sonalksis have lost the plot! But if you can be arsed to spend a few hours downloading this, giving away your personal information and risking f**king up your system, then go for it. It's a good plug in.


The meter levels in your DAW of choice. They are all different. For example, EnergyXT2 does not allow you to see the values, so you are groping in the dark.

Buy VUMT. Download the trial of EBULoudness. Compare this to the metering in your DAW.

Make your own notes. Something that will make sense to you later. Use your own music. If it tends to be bass heavy then you will have to lean towards that side with your metering (RMS is important here). If it is sparse but with lots of drums, then that will take it the other way (Peak is important here). And compare all this with reference tracks that you like and see how they measure up with the peak/rms scale of things.

[Compyfox]

EBU R-128 meters only make sense for mastering.
At least IMO.

For tracking and mixing, host built in digital meters and maybe a third party VU like Klanghelm VUMT or Hornet Audio VU Meter and you're pretty much set.

[codec_spurt]

EBU R-128 meters only make sense for mastering.
At least IMO.

For tracking and mixing, host built in digital meters and maybe a third party VU like Klanghelm VUMT or Hornet Audio VU Meter and you're pretty much set.

Exactly. Learn this. But don't over think it.

I certainly do not measure every single signal that comes into my DAW.

I have my faders down to -12dB and this will afford me the head room necessary and not degrade signal quality at 24/32bit.

Buy the Klanghelm VUMT - it is a multi purpose unit, and only costs a fiver. It opens up a whole new world with metering you can trust.

If you simply can not afford that then have a look at http://sleepytimedsp.com/software/str-bundle/



It's free. But you still want the Klanghelm VUMT on top. Then you are sorted.

[Compyfox]

VUMT now costs 8EUR, Hornet Audio VU costs 4EUR, SleepyTime DSP VU is free.

All have their strengths and also their weeknesses.

VUMT (Klanghelm) does offer a VU and a QPPM, +/-24dB gain, needle style meter, ballistics can be setup
Stereo Channel (SleepyTime DSP) is a bit more than a VU meter, +/-24dB, needle style meter, ballistics can't be setup IIRC (haven't used this one in a while)
VU Meter (Hornet) is just a VU, but comes with an auto-gain mechanism (that I've yet have to test), but the boost is only +15dB, is a horizontal bargraph, no color codes possible


Mentioning an EBU R-128 meter is good and fine, but this one is for specific tasks that have no real use while tracking/mixing. It's also really confusing for those that are new to learning metering tools properly. So let's focus on the basics first please. And that is a VU and a digital meter.

Really, more than enough for the time being.