Anti-Logic Studio

Mixing / Mastering / Editing / Restauration

How much headroom for mastering ? ...and what is it really anyways ?

Search other topics here...
 

1. Headroom ? What is this about ? *head scratch*

Often times you will read to leave the amount of 6db headroom in an audio file which you want to export and send to mastering. In our modern times that number 6 though is not set in stone and really not all that important too. Mastering requirements could very well also tell you to leave 9db of headrom or 12db or, to go the opposite route, to leave only 3db of headroom. They could also not even specifiy an amount but rather say "no clipping".


Bear with me here, I'm never giving answers like "Simply do this and don't worry, it will be fine!". I want to you understand the matter and I'm trying to give you insight, so that you could always make your own conclusions or even disagree with me, if my article were suffering from a thinking error. We're all humans here, maybe, just maayybee reptiles.
#skepticalhippoeyes

If you want to understand why you are reading about headroom at all and of different amounts, here is the principle behind it: In your exported audio file, surely you want a perfect reproduction of what you were mixing/hearing out of your DAW, right ? Darn right ! As a consequence, you never want to process a file that is clipping (peaking above 0db) because it can sound distorted.

2. What's the big deal ?

Clipping causes what's called nonlinear distortion. This happens when parts of the signal, those that go above the maximum that a digital device can handle before distorting, let's call that the 0db marker, are "chopped off"...or "clipped off"...hence, Clipping.
   So imagine a sine waveform with an increasing amplitude, it reaches towards 0, eventually hits it and goes even higher. When it does that, the peaks will go from a round shape to a square one as they are "chopped off".

 

 So look at the hard clipping to get a clue what digital clipping is.  But ey, know this: In an analog devices nowadays, driving a signal above this point will cause a different kind of distortion, one that can actually enhance the sound, because analog devices today have components inside that start a sort of early limiting in dependence of the intensity of the level. It's called "soft clipping". So with soft clipping you don't get that harsh digital clipping but a softer version and if you wanted to visualize it, imagine it as softer edges.

So look at the hard clipping to get a clue what digital clipping is.

But ey, know this: In an analog devices nowadays, driving a signal above this point will cause a different kind of distortion, one that can actually enhance the sound, because analog devices today have components inside that start a sort of early limiting in dependence of the intensity of the level. It's called "soft clipping". So with soft clipping you don't get that harsh digital clipping but a softer version and if you wanted to visualize it, imagine it as softer edges.


Now, I know you are asking "who's doing all the chopping then?". Of course there is no digital scissors, sword, axe, dagger, not even a guillotine there that does it. Rather imagine it as a waveform (again for better understanding think of a simple sine wave) whose positive(1) and negative(-1) peaks simply exceed the capacity that a device can handle, that's it.

Back to the headroom and back to our principle, you never want to process a file that is clipping. That should be one of a few of every audio engineer's maximes. Dare I say, it should be one of those principles of life - only use quality material for further processing in order to achieve a quality result.
   I have mentioned it before and it is still true, mastering is the last step in a chain of steps to create and finalize a piece of art that is ready to hit the market or a target group. Why ruin it by not sending the optimum into mastering ?
 
So while it should be obvious and understood that you must not send a clipping file to mastering, we hand out the requirement to leave a specified (or unspecified) amount of decibel of headroom to prevent the mastering engineer from running into problems.
   First of all, if clipping in audio material could be heard before mastering, of course it would be heard after mastering!!! Congratulation, the master is ruined before it was done. Even if (and that's a BIG BIG IF) the engineer would or could try to go deep and the clipping parts, that's not how you treat and respect that final step of polish and quality control. Ultimately it's not how you treat your art or the consumers of that art. 

You don't eat excrement, you don't drink fuel, you don't disrespect your mother and even if you're a complete rebel, you sure as hell don't expect mastering to deal with clipping !
*unless you make the engineer a millionaire, then by all mean, please go ahead :)

3. Tell me already, how do I do it ?

Audio material that is exported with headroom will in the ensure that no clipping happened before mastering. Then, the good mastering engineer will ensure to export a product without clipping. That's the theory !
   In practice you have to make sure that neither your master channels is driven into red, but also that none of your plugins are clipping internally. Your EQs, your synths, your effect plugins, they all better not be clipping.

4. So is it enough if I send a file that is peaking at -1dbFS, so with just 1db of headroom ?

In theory, yes ! Should be fine because the mastering can adjust the level of the material before it goes into his mastering chain.
    But preventing clipping is not the only reason for having headroom. In mastering there is a good posibility that we alter the sound by adding or taking away from the frequency spectrum, which will alter the level of the material before it hits the limiter, our Big Loudness Changer and "boarder control". Had the engineer used a file with -1db headroom in his mastering chain, he'd ran such a hot signal to his device/s that any boost by an equalizer or exciter or saturator or anything else, could cause clipping. A good amount of headroom is actually supposed to leave a comfortable amount of room between point "is" and point "to be", for the processing that might happen in between them.

5. So why not settle on a reference amount, a global standard for headroom so to say ?

Because engineers have their own preferences and workflow. I can only tell you about my own preferred headroom but we can't make everybody work in the same way.

I ask producers or mixers to submit audio files that peak at -6db full scale. Let's remember what that means: It is a delta of 6 specific units (namely db or decibel) between point a = -6dbfs and the absolute upper limit of b = 0dbfs ...ergo, 6db headroom between a and b. 

In case you are wondering about the term dbfs, I only mention it because I can hear how somebody would say "oh you have to write dbFS in that context, else your article is full of BS". To our regret, those sticklers exist and technically it is the right unit to use, so !

dBFS is a scale that is used for digital signal chain metering and  refers to an absolute scale, ranging from some point in the far negative (-n dbFS) to a maximum of 0dbfs. There are other scales and you might know this one: dBSPL for sound pressure levels (street noise, plane, voice, etc). There are many more for different purposes like dealing with signal flow and voltage, with radio technology, lots of different absolute or relative scales. 

6db of headroom it is for me ! But in reality I rarely ever get a file like that :)
   Levels are often all over the place, especially in songs for a compilation when you are dealing with different producers/mixers. But they come close enough often times. When you see your master channel meter in your DAW peaking at -6,2db or -5,8db, there is no need to adjust the master fader for some 0,2db there. Export that stuff and done. Good enough for me !!! 
I will then go and adjust the level to my exact preference.
   I actually use a trick to do it and also ensure consistency with peak levels of all the songs that I get for mastering. I want all songs that go into my chain to peak at exactly -6db because my mastering chain preset, from wich i start, is finetuned to that peak level and I go on from there.

6. How I adjust all material to a peak level of -6dbFS before mastering ?

It happens in a step that I call pre-editing. I load the file into a wave editor and check the real word length to find out if a song has actually been exported at 24bit or only been upscaled to 24bit from a former 16bit. I also check things like the the samplerate or the beginning and end of a song and I will normalize the song to my preferred peak level of -6dbFS with the "normalization" feature, which brings the peak level of the material up or down (dependent on where it was before) to the exact value of my choice.

There is another BIG advantage to this: Consistency in case you have to receive a second version.
   
Imagine you master a song and you agree with the client to get sent a second version with slight mixing changes. You have normalized version 1 to have a headroom of your choice and you can now normalize version 2 to have the same headroom. Now both will be very very similar in level and you can load up your old mastering project, insert version 2 and go from there without having to start from zero or ideally without having to make big changes.
   Doing this is very important for me - and I speak from huge experience here - because when I received a version 1 that met all your mastering requirements perfectly, that doesn't always mean the version 2 will be anywhere your requirements. I have seen it all by now and sometimes we are dealing with a producer duo where person 1 exports this way and person 2 exports another way, there you go, congratulations ! Or sometimes you are dealing with freaks, stoners, weird geniuses and often with artists - which that says it all :D

Questions ? Hit me up at anti.logic@gmx.de

Until next time
Yours truly

 
Search other topics here...