Benedict Roff-MarshThis article looks at all the main issues around understanding levels in mixing and mastering.
We look at what the numbers, like dB, RMS, VU, LUFS mean, how to use the numbers, and when not use the numbers as, after all, numbers are not art.
The companion video is essentially the same material. We encourage you to take in both as this will help you to understand the deeper concepts here.
What are levels and why do we care?
In mixing groups everywhere there are endless questions indicating confusion about volume, gain, and levels and what those numbers need to be to have hits.
Levels are how we measure and monitor (keep track of) how loud our music is, and appears to be, at any point in time. This matters for several reasons, of which masters are only one small part.
Let’s start by looking at what levels really are by understanding what they are not.
Level vs. volume
As humans we don’t see and hear the world exactly as it is. We perceive our world around us. This means that information arrives at our senses and we “put it all together” in our brain.
How one person puts that information together can result in the perception of a “white cat quietly meowing” whilst another perceives a “purple cat yowling at titanic proportions”.
Machines however see exactly what is. Note how the cat perceptions were both emotive. A machine would see “cat, tortoiseshell, part tabby, articulating at -24 dB moving to -18 dB”. That is a quite a different thing.
Here’s another way of seeing that:
Here are two people. Are they the same size?
I assume that you say no, they are not the same size; look that one is small and that one is large.
If I then add this extra information:
Now, with the new information, you probably realize that they are likely the same size. We mis-perceived the one on the left to be a small person when in reality they are a long way away from us and perspective makes them appear small.
That was so much fun, let’s do another:
Who here is larger? Who is standing in front of the other?
The person on the left appears larger as they stand above – seem taller. The person on the right seems to be standing in front.
There are two reasons for this:
- the indication of perspective makes us believe the person on the right to be forward,
- standing outside of the frame, makes them appear to “leap out”.
Yet they still seem smaller despite being the same exact size. Perspective again has our brain adjust the person – at a greater distance – to be larger to make sense of their appearing to be the same size. Kooky!
Now you start to see what mixing is really about and why levels are expressed as numbers that are presented in seemingly odd ways.
Measuring audio levels
There are several terms that you will met in audio level matters:
dB = decibel 1/10 of a bel
The decibel (small d large B) is the standard measure in audio. It is delightful to know that it is a tenth of a larger unit but somewhat pointless in most practical ways. It just makes me look amazing to quote technical things.
Peak = exact level
This is the easy one as like the cat observing machine above, peak shows exactly what the machine hears at that exact moment in time. Useful but not the most useful at all.
PPM = peak program meter
Is a slightly averaged form of the above, usually about 10 ms. Rarely used a lot for music mixing. Feel free to ignore most the time.
VU = volume unit
Is a level measurement based not on what is there right now but how we perceive level in our feelings. We average everything.
Volume units read very differently from peak, e.g., a snare can show a peak of -3 dB but VU of -20 dB. The machine sees the -3 dB transient; we are vaguely aware of it (as you can tell if it is chopped off) but overall we feel that the emotional event of that snare hit was -20 dB.
Officially VU are measured as VU not dB, but for the sake of ease I used dB as that is what most meters appear to show.
Again, remember that if we are emotionally excited about that snare, we perceive it to be larger than if we are completely disinterested.
In the latter case our person’s VU on that event might be zzzppffft, otherwise known as -104 dB.
Commonly the form of meters that use a needle that “flicks” as noise gets loud are all called VU. This is not always accurate either as some read more like peak than actual VU, especially the software type. The Ballistics – as in how slowly they move gives some sense of VU or…
RMS = root mean square
Is another averaged form of level over time. There is complex math here we do not need to know except that RMS and VU are commonly used as synonyms for each other. RMS simply means what we perceive.
LUFS = loudness units full scale
Is the new kid on the block being used by content providers like Nettlefix and Spotifry to TRY to make content creators behave, i.e., not hammer content that is MUVVA EFFING LOUD at ordinary people hoping not to spit their soup all over the place when the next song or advert comes on.
Like VU, LUFS attempts to provide a unit of perceived volume rather than just the facts ma’am level. More on that ball of wax later.
Why scales?
From the info above we can see that level metering is a bit of a squirrely one. These units of measurement operate within a scale, just like notes operate in a scale like C major. The numbers only make sense in the context in which you are wanting that information.
If you want to know exactly how much headroom (space before you hit the top of the can) you have at any given time, you will look at a peak meter.
Even better if it is a Peak Hold Meter that holds the peak value for either a few seconds or until you reset it. If the peak for the whole song is -0.5 dB then you are good to go as the top of the can has not been hit.
If you want to know how people may be feeling about the overall level of your song, you read RMS or VU values, which let you know how they will feel about the overall sense of volume of this track.
If you make the mistake of trying to get your RMS or VU to be the same as that -0.5 dB peak, while your listeners might feel your track is huge for a few seconds, they will start turning their volume knob down or leaving.
I have several records I find very hard to enjoy because, while there is nothing wrong with the peak levels, the RM or VU numbers are so high I feel like I am being pushed by a bulldozer.
And I don’t mean that in a good way, but like I am being crushed and unable to breathe. Not a good plan for making a fan.
What those records lack is dynamics, which is the difference between the LOUD and soft bits. Some silly sausage figured that if everything was all-slammin’ all-the-time the record would be Epic! Bad plan.
The solution there is in the “Pictures of Matchstick Men” above (tee hee – YouTube it). It is all about perception. The absolute best way to make the right decisions about where levels should be in your song is to use your senses.
But if you haven’t developed those senses, or worse, still have had then mis-developed by only listening to Metallica “Death Magnetic”, Judas Priest “Firepower” or just about anything on Spotifry in the last 20 years, then bring in a truly skilled mix engineer who understands those pictures above in music terms.
Otherwise learn the value of crest factor.
Crest factor
Crest factor is the difference between the peak and VU values. The image here shows the peak at 0 dB whilst the VU is also reading 0 dB.
This tells us that both the machine and person will probably be happy enough – not feeling unsatisfied or hammered.
Depending on your system the 0 dB VU can be opposite -18 dB or another figure on your meters. Reason (and I) like -12 dB for 0 VU. That is a very nice compromise for a very natural mix. Note, I say natural, not overloud.
This metering configuration means that when I bring my song up to the masters and have the peaks tickling the underparts of 0 dB and the VU gliding gently around the -12 dB (0 VU), I have a crest factor of 12 dB.
A 12 dB crest factor (simply) means that I have a dynamic range of 12 dB, which is emotionally like a really nice sense of breathing space inside the mix between the peaky bits and the average bits.
Less breathing space than that and people start to feel “imprisoned” or “bludgeoned”, which is rarely the outcome you want.
Especially seeing that you want/need people to want to come back to your song/s. How many people go back to prison for fun?
While I get many of you are saying “oh butz I needz to be like competitive dude like wid da loudnezz on da streams man. dey like all up at – 0.00004789 LUFS. Not my fault gotta be dere brah.“
Ok, your call but my take is that you lose more people who would become real fans (people who pay you) with overloud than with the one person who winds his volume knob in the upward direction to get more of you.
Matter of fact, a mix that your listener wants to embiggen – so it is even more immersive – is exactly what you want. A song that people don’t want to interact with (past turning down/off) is not a good thing.
Gain staging
Gain staging became the hot topic for a while there in DAWville. There were a lot of strange things said and done. Things that still have people confused where there should be no confusion.
Staging means setting things up. You know, like putting the drum kit and guitars on the stage so they are there when Iron Maiden come out to play. Otherwise, things may not go to plan. That is staging. Gain staging is the same thing only for levels.
The image above shows a signal going through three (very Retrowave) processes before hitting the masters. Note how at each step the signal gets larger. All fine until it hits the masters whereupon that signal is too big to fit in to the box. That is problematic.
We need to turn that level down before it will go in the delivery box. We need to grab a… no, not a limiter, no, not a clipper, not insert plugins here. We need to adjust gain with a Gain or Amplifier Trim to reduce the level so that signal will slide into its delivery box.
Gain staging is: preparing the signal for what comes next.
You will read a lot more guff about having to have every signal at -18 dB all the time but that is BS. Ask Andrew Scheps, (apart from “Death Magnetic”) he knows what he is doing. He talks about how he adjusts level/gain for a signal only if it needs changing.
Here’s an example: Dade Mustang walks into the studio with his DC Rick Poorlock guitar and plugs it into the Megga Metal pedal on the floor.
Dade starts playing but it is not very metal. Not even slightly metal. The problem is that Musty Dade forgot to turn up the volume knob on his guitar.
This means that the Megga Metal pedal wasn’t getting enough level to be even slightly Megga. Andrew strolls into the live room, winds that knob to full and gets Dade playing again. Now it’s metal. Gain is staged.
Gain staging is simply making sure that the levels at that point are appropriate for what needs to be done. No more. Definitely no being obsessive.
It does make sense to start a mix with enough headroom to allow you to mix. If you start with every individual sound hitting the meters at +6 dB, by the time you have even two sounds in, your masters are at +12 dB which is a lot to shed later.
Matter of fact, even four sounds in that mix can have the masters at +24 dB, which all but makes decisions impossible.
This is why there are suggestions to start the mix with every track at -12 dB so you aren’t constantly struggling to bring the Master Fader down (which I never ever, like totally ever, do during a mix).
To do that you simply use the Gain knob at the top of the Channels to “stage” your mixer ready for mixing.
Why I think LUFS and Streaming Service Leveling do not matter
Which brings me back to LUFS the new(ish) standard that is all the twitter when people obsess over how loud their songs can or should be. See above, “should” is dependent entirely on the material – trust that please.
If your meters read low but the scene and story of the song feel great then you have it.
If they still feel like they need another 37 dB pumped into the Limiters before the track might almost sound exciting, then your mix is broken. Maybe even the composition and performance.
LUFS is an interesting idea in that due to Loudness Wars and other mean practices by those unsure that their work is good enough, it was felt that a standard would be a great plan to ensure that perception of volume from one program to another would not cause people to feel assaulted by one song then lost with another.
The thing is that this was handled by AM Radio in the 60s (well before, probably). Handled by FM Radio in the 80s. Sure TV was less kind in that they let adverts be LOUDER than the programs, but that was down to the politics of payment more than any technical or technological failing.
So… why do we need new numbers that don’t remotely change how people perceive music?
If you worry about having your tracks turned down… don’t. Logic alone says that if your tune is de-volumated, it was too loud for the enjoyability of the listener hoping to be entertained by your blatty beatz.
If you did come in a bit softly, the level will go up a bit to put in you in da game. If however you feel that the moment that 1.5 dB is shaved off your song it dies, then you are saying that you know that your song only works when it is at Boeing levels (120 dB – pain threshold). This is a worry.
Roxy Music “Avalon” works at any and every (practical) volume that any (remotely sane) listener has their stereo placed at. Your song should too.
Sure, some styles are less flexible, but that is part of your job to work out how to Translate. If your music, like my Space Music, can be a bit diffuse for a tinny trannie, then simply don’t aim for people listening on their tractor. Swap to Country if tractor drivers are your peeps.
Final words
Levels are simply ways to translate numbers from machines to our brains. The numbers must NEVER be more important than the senses.
I have been amazed at some records that I have pulled up into my DAW and realized how quiet they are technically, when emotionally they are like monoliths. This is because the people involved in those records were using their senses and feelings far more than numbers.
More importantly those people trusted their story and that is what really translates to the listener. You can’t put that kind of trust under a meter, but you sure know when it is not there.
Listen to a record from a truly great singer like Sinatra, Rod Stewart, Dio, or Madonna and compare to some Desperate Des dying to get a date. Note that Des exudes fear. This is why he spends all his time talking about levels and gear while Rod The Mod talks to blondes.
Move past using only peak metering ASAP and use combined numbers which most decent DAWs offer on the mains so you can see the important things easily.
Further, don’t use boxes that list enough numbers to start your own telephone service. Use your senses. If you don’t feel it, either train yourself, or hire an experienced mix engineer (who may even help train you if you are wise in how you deal with them).
Remember that many of the truly great records were made with only VU meters. Peak meters arrived pretty late in the game.
Even then, the first peak indicators were just a LED that turned on over 0 dB. Pretty LED meters were not common in most studios until the mid-80’s.
All those great records were made with feel and courage. Numbers are useful but not at all what makes a song great.
