Metering in digital domain

offcenter2005

New member
levels in digital domain

I have been reading a lot of info on where to keep levels when recording and mixing strictly ITB. Almost every thread i have been reading ends up in a debate and the main reason for the thread gets lost to an argument. When I first started recording digitally i would keep the input levels as close to unity as possible and my mixes sounded like crap. Then I started to crank my monitors and record and mix with the levels significantly lower. The mixes started to sound better but i still think that my gain staging thru plugins are wrong. I know this has been discussed a lot but could someone give a little advice or point me to a good thread that has good info and less arguing? Thanks in advance. Id also like to add that ive been using plugins like ssl comp and v-comp, from what i understand there are specific gain staging recommendations for these plugs to respond as intended. I also have been using the Sonalksis free g to check the levels of every channel.
 
Last edited:
This is an old chestnut that even experienced engineers have trouble with. I'll try my best.

I think the first thing to get your mind around is decibels, their suffixes, and the difference between them. For instance, most analogue gear is metered on the dBu (decibels unloaded) scale and this can be confusing when trying to relate it to the dBfs (decibels fullscale = 0dBfs). The first thing to notice is that 0dBfs does not equal 0dBu. On an analogue console you'll most likely see 0dBu somewhere in the middle of the meter with a positive scale all the way up to +22dBu or more. This scale was devised to indicate headroom above nominal operating level, which is supposed to be around 0dBu to +4dBu. Conveniently, a 1kHz sine tone at +4dBu will reflect as 0VU on a VU meter (Volume Units), the slow-moving, loudness-type meter of old. So, how do we translate this over to the digital system?

Well, the first thing to know is that all digital gear that converts an analogue signal to digital will have it's own internal calibration. This calibration will translate to 0VU/+4dBu/dBfs. Thus, this is a typical calibration equation:

0VU = +4dBu = -18dBfs

This equation tells us that in this specific system, it's a good idea to keep your peaks around -18dBfs to leave sufficient headroom for erroneous levels. Now, every system is different and every interface has it's own calibration so if I were you I would check online for the calibration of your particular interface. The dBfs figure will give you an indication of where to keep your highest peak levels.

Of course, there are no hard and fast rules, so use your discretion. It's OK if a few peaks go over -18dBfs. That's what the headroom is there for. As always, however, conservative levels are a better idea than slamming anything.

Now, when it comes to plugins, the basic rule is to keep perceived loudness equal at the input and output of the processor. This usually renders peak level useless to reference. You have to use your ears. It's not that difficult, however, so make good use of the bypass button! That's pretty much it.

Hope that helps.

Cheers :)
 
Thanks for the replies. From what i have read , in between all the BS thats what I assumed was the underlying message. I started noticing a fuller clearer mix when keeping my levels lower. If i turned up my monitors and turned down the channels everything seemed to come into focus. Thanks again.
 
I started noticing a fuller clearer mix when keeping my levels lower. If i turned up my monitors and turned down the channels everything seemed to come into focus.

Most modern DAW software uses 32-bit (or 64-bit) floating point math, so signal levels within the software have very little effect on sound quality. For proof, see the article I wrote for GC Pro's Audio Solutions magazine:

Audio Solutions magazine

Click Archives, then select the Summer 2011 issue. It explains DAW internal calculations in detail. I also prove the same points with nulls tests in my hour-long AES Audio Myths video in the section starting at 53:39. Note that some distortion and tape-sim type plug-ins are level sensitive, but most "normal" plug-ins are not.

Now, getting audio in and out of a converter is another matter, though again any competent converter should be able to handle a wide range of input levels without compromising the sound quality. For example, look at the specs on the last page of this Lavry converter manual:

http://lavryengineering.com/pdfs/lavry-da924-manual.pdf

Even when sending a loud signal at 1 dB below clipping, the distortion is only .0009 percent. Other converters may not be this clean, but most are still very clean right up to the point of gross distortion. I believe that when people reduce levels and notice the sound improved, the problem was occurring somewhere else in the signal chain.

--Ethan
 
Most modern DAW software uses 32-bit (or 64-bit) floating point math, so signal levels within the software have very little effect on sound quality. For proof, see the article I wrote for GC Pro's Audio Solutions magazine:

Audio Solutions magazine

Click Archives, then select the Summer 2011 issue. It explains DAW internal calculations in detail. I also prove the same points with nulls tests in my hour-long AES Audio Myths video in the section starting at 53:39. Note that some distortion and tape-sim type plug-ins are level sensitive, but most "normal" plug-ins are not.

Now, getting audio in and out of a converter is another matter, though again any competent converter should be able to handle a wide range of input levels without compromising the sound quality. For example, look at the specs on the last page of this Lavry converter manual:

http://lavryengineering.com/pdfs/lavry-da924-manual.pdf

Even when sending a loud signal at 1 dB below clipping, the distortion is only .0009 percent. Other converters may not be this clean, but most are still very clean right up to the point of gross distortion. I believe that when people reduce levels and notice the sound improved, the problem was occurring somewhere else in the signal chain.

--Ethan
I think a big part of the sound changing and sounding more focused was the fact that instead of changing the levels with the fader I changed the trim level. After that i readjusted my plugin chain settings. The real problem probably came from the gain staging through the plugins. Maybe im wrong but if I were hitting the plugins too hard with the signal it was causing some distortion.

Are you saying that all signals coming into the converters should be showing between -18 to -12 for optimal sound and trimming will do nothing but lower the already recorded signals volume?
 
No, what I'm saying is most plug-ins sound exactly the same no matter what level you throw at them. Now, this is not true for all plug-ins, but it is for most of them. This is very easy to test for yourself, and I urge you to do that. It won't take long, and you'll learn something very important about how your audio software works. See my article and video linked above for more info on how to test this. I promise it's well worth your time. At least if you want to know the how and why aspects of audio.

--Ethan
 
When I first started recording digitally i would keep the input levels as close to unity as possible

What do you mean by "levels as close to unity as possible"? Unity applies to gain, not level. Using terms incorrectly like this could contribute to misunderstandings.
 
What do you mean by "levels as close to unity as possible"? Unity applies to gain, not level. Using terms incorrectly like this could contribute to misunderstandings.

What i meant was when setting the gain of the input and output signal to the mixer-converter-daw, i would keep the meter reading at near unity gain = 0db. Is this not the right terminology? What i meant was input gain.
 
What i meant was when setting the gain of the input and output signal to the mixer-converter-daw, i would keep the meter reading at near unity gain = 0db. Is this not the right terminology? What i meant was input gain.

What the meter displays, for example 0dBVU, is the level, not gain. The input preamp adds gain as needed to make the signal the desired level. Where you get a unity gain setting is on the faders where it is the 0dB point. It just means the fader isn't adding or subtracting any gain and the signal passes with its level unchanged.

I would say you're setting the input gain for a level of 0dBVU on your meter. That probably comes out as +4dBu at the outputs to the converters. Depending on the converters that may end up in the vicinity of -18 to -12dBFS.
 
What the meter displays, for example 0dBVU, is the level, not gain. The input preamp adds gain as needed to make the signal the desired level. Where you get a unity gain setting is on the faders where it is the 0dB point. It just means the fader isn't adding or subtracting any gain and the signal passes with its level unchanged.

I would say you're setting the input gain for a level of 0dBVU on your meter. That probably comes out as +4dBu at the outputs to the converters. Depending on the converters that may end up in the vicinity of -18 to -12dBFS.

If i adjust the converters preamp level that would boost the signal above the -18 to -12dbfs then right? I think the preamps in the allen heath mixer that use has less noisy pres so i adjust the gain with the mixer preamp and leave the converter pres down, also a lot say that recording with a level of -18 to -12dbfs is pretty much where it needs to be for digital recording, would that sound to be true in this situation?
 
No, what I'm saying is most plug-ins sound exactly the same no matter what level you throw at them. Now, this is not true for all plug-ins, but it is for most of them. This is very easy to test for yourself, and I urge you to do that. It won't take long, and you'll learn something very important about how your audio software works. See my article and video linked above for more info on how to test this. I promise it's well worth your time. At least if you want to know the how and why aspects of audio.

--Ethan

So how do we know which plugins do and which plugins don't? Should we test each one? Logic, then, would dictate that it's better to just practice conservative levels.

A little FYI, if I may.

Just so everyone knows, Ethan prides himself on being an audio myth debunker. To my knowledge he's never made a hit record and the things he says pisses off a lot of the the people that do, because they vehemently disagree with him on much of his "debunking".

I can list at least 5 or 6 of them off the top of my head, and these are people ranging from having 50 years in the business, to top 10 hit maker engineer/producers. There are epic threads all over the internet forums where Ethan and his adversaries battle out these points of contention.

My opinions are somewhere in the middle.

Cheers :)
 
So how do we know which plugins do and which plugins don't? Should we test each one? Logic, then, would dictate that it's better to just practice conservative levels.

Anytime you are using a plugin that is an emulaton of real gear and the way it acts there is going to be a simulated 0VU point well below 0dBFS to allow you to push into the simulated headroom to get that distortion that comes from pushing levels into gear

Many plugin manufacturers use -18dBFS as simulated 0VU. Waves emulations do although it is user definable if you want to change as do UAD with the exception of their tape sims Studer 800 and Ampeg ATR 102 which use -12, satson, SK note, IK etc all have emulation plugs and this concept of simulated 0VU and simulated headroom and distortion is now very common in plugins

So if you are using any emulation plug anywhere in the chain you are going to need to be aware of the levels going into that plug to get the result you want, regardless of what the recorded levels were

Also, while converters may be clean all the way up to 0dBFS, if you are using any analog equipment in your recording or mixing chain it will usually be looking for a signal of 0VU based on a voltage measurement so if you are going to 0dBFS on your digital meters you could be pushing levels of +22dBU into that gear which could make it sound pretty bad

I like the master meter in REAPER which allows you to define where 0VU is. it will still give you the dBFS scale as well but you now have a reference as to where 0VU levels should be. Satson's Sonimus metering is also useful for this at the track level

Long story short, if you shoot for levels of around -18dBFS (RMS not peak) you will never go wrong and you can always bring the finished mix up to more commercial levels at the final stage once the two track mix is finished
 
Yeah, I have neve heard of this simulated 0VU as you call it. That's because plugins use floating or fixed point binary math internally. I'm going to have to go ahead and call BS on that one.

Translating 0dBfs to dBu depends on the calibration of the hardware. For instance, my Aurora 8 clips at +20 dBu. This means that when the daw is outputting 0dBfs, I will read a level of +20 dBu at the outputs of the converters.

Regarding plugins, I doubt this simulated 0 VU. Has anyone else ever heard of this? I stand here willing to be corrected. As far as i understand it, the internal calculations are all carried out in binary numbers in what amounts to base-2 arithmetic. In floating point math, values representing the positive and negative phases of the wave form are represented by a positive or negative integer between 1 and -1, which is full scale (0dBfs). Levels greater than full scale can also be temporarily calculated in floating point math but at that point the code must be properly written so that he result is less than 1 or more than -1 at the output. Some plugins do it well, some add distortion and some just crash. This, of course, all depends on the proficiency of the developer. On top of that plugins will often have internal precisions greater than the recorded audio and will often dither at the output. This is another area where distortion. An creep in.

All plugins are not created equal.

Plugin guys, if I got anything wrong please correct me. I am not a programmer.

Cheers :)
 
Sorry Bristol poss. Didn't see you were talking about saturation emulations. In that case it make sense. Damn phone foruming!

I retract my criticisms of your post.

Cheers :)
 
Regarding plugins, I doubt this simulated 0 VU. Has anyone else ever heard of this? I stand here willing to be corrected.

From the waves Kramer Master tape manual:
features a modeled analog VU meter, where 0 dBVU = 1.23Volts RMS = +4 dBu at 1 kHz. Using a 700 Hz tone at -18 dBFS, input and output levels are equal. The default VU meter calibration is -18 dBFS = 0 dBVU, which we found to be optimal for achieving the desired sound when the meter action hovers around 0 dBVU. For hot digital signals peaking close to 0 dBFS, this will require lowering your Record Level proportionately to achieve “proper” tape sound.
From the waves SSL bundle:
3. An Input Trim Button enables you to trim the input to the channel by ± 18db. The
plug-in is aligned so that -18 dBFS = 0.
From the UAD System manual
Operating Levels
Except as noted in Table 4 below, the internal operating level of all UAD Powered
Plug-Ins is –18 dBFS. 0 dBFS is calibrated to +4 dBu with 18 dB of headroom,
so 0 dBFS is the equivalent of +22 dBu in the analog domain.

and so on
 
From the waves Kramer Master tape manual:

From the waves SSL bundle:

From the UAD System manual


and so on

The tape saturation plugins were the ones I was referring to. I dont know any of the math behind digital recording or much of the technical side but that is why these forums help. I have been getting the mixes to sound better and better the more I learn about this stuff. It really seems like little mistakes that I make ruin my mixes. Although I can add that when im mixing I use my ears more than anything. But I really think there is a technical side to this that cant be overlooked.
 
So how do we know which plugins do and which plugins don't? Should we test each one?

Yes, unless you're happy to remain ignorant about how the tools you use actually work. The same applies to outboard gear too. Push it and see what happens. Maybe you'll like it, maybe you won't, but at least you'll know which end is up.

Just so everyone knows, Ethan prides himself on being an audio myth debunker. To my knowledge he's never made a hit record and the things he says pisses off a lot of the the people that do, because they vehemently disagree with him on much of his "debunking".

Ah, another hater intent on discrediting Ethan. Nice. Didn't you hear the war is over? Regardless, maybe 1.5 Million views on YouTube and elsewhere doesn't count for much these days, but I'd call that a hit. Most views:

A Cello Rondo

Newer render with much better quality:

A Cello Rondo - HD Version

I was designing audio gear and producing music for national jingles and soundtracks while you were still in diapers:

Ethan's Audio and Music Bio Page

And the list of your hit records can be found where? Not that how many "hit records" one has produced affects the validity of their arguments. But you brought it up.

if I got anything wrong please correct me. I am not a programmer.

No kidding. As it happens, I am a programmer. Look, whoever you are, if you don't have anything to offer but insults and accusations, you could avoid embarrassing yourself by just staying out of it. The people I "piss off" are those who hold strong opinions, but lack the knowledge and foundation to understand the science or express themselves intelligently. So all they have left is insults. Sound familiar?

You would do well to read the article I linked to previously, and try to understand it. For extra credit, read and understand my Perception article, and watch my hour-long AES Audio Myths video which proves these points. You'll come out knowing a lot more than you know now. And if you really want to understand audio and hearing, buy my book The Audio Expert and read it all the way through.

--Ethan
 
Many plugin manufacturers use -18dBFS as simulated 0VU ... if you shoot for levels of around -18dBFS (RMS not peak) you will never go wrong and you can always bring the finished mix up to more commercial levels at the final stage once the two track mix is finished

Yes, but understand that the concept of RMS and average levels derives from the need to know how loud something sounds. And when used for setting levels, it's really only appropriate for analog tape recorders. With digital systems, the audio is perfectly clean right up to the point of hard clipping. So all that really matters with DAW software is keeping the peak levels safely below Digital Zero. If you can hit 0 dBFS without distorting anything in the analog path, or being so far down the analog noise floor is a factor, record levels don't matter. The notion that DAW software has a sweet spot is definitely a myth, though - again - some plug-ins add intentional analog "flavor" that increases at higher levels. Other 32-bit plug-ins are not affected.

--Ethan
 
If i adjust the converters preamp level that would boost the signal above the -18 to -12dbfs then right? I think the preamps in the allen heath mixer that use has less noisy pres so i adjust the gain with the mixer preamp and leave the converter pres down, also a lot say that recording with a level of -18 to -12dbfs is pretty much where it needs to be for digital recording, would that sound to be true in this situation?

I don't know what converters you have so I don't have any idea what the correct settings would be. I take it you're going from an A&H mixer into an interface, perhaps the ADA800 on the gear list in your sig line? Are you using the TRS line inputs? If so then +4 is probably where you want to set the knob (since presumably you're sending it a +4dBu signal). If you're using the mic inputs there may be too much gain even with the knob all the way down.
 
Back
Top