How much headroom should I shoot for on rhythm tracks?

MG-Chris

80's Thrash > *
Nov 28, 2010
108
0
16
Boston, MA
www.metalguitarist.org
Bit of a newbie question, but I searched around and found about 900 other awesome threads but nothing addressing this. (I'm sure it's covered, just not enough coffee on my end..) So with my apologies for asking something that's probably been asked 300 times..

I'm about to start tracking rhythms for my album, and I'm running an Axe-II along with a mic'd 6505+ into my DAW for it. How much headroom should I be allowing on my raw rhythm tracks before sending them to the master mix? I'm using Superior for drums and a mix of Trilian and my real (sub-par) bass playing for bass, so the only real levels that I need to get down beforehand are my guitar tracks.

Is 6db enough, or will I run into clipping once I post-eq and such? I'm playing a lot of high gain stuff/shred/power metal, and I usually try to get my max input level by just slamming the shit out of the biggest, lowest (loudest) chord that I can a few times and seeing where it peaks, and then going from there.
 
Ultimately, the only thing that matters is to not clip your converters. Although, the most beneficial for you would be to adjust levels, so that every single track peaks between -6 and -12 dbFS (or, around -18db RMS). Master should peak in similar range, but NOT OVER -3 dbFS in any case. Generally, when you apply plugins, you should try and match their inputs/outputs every time, so that plugin only work as intended, not to boost any levels with them (it's ok if you do that here and there, just don't make it your standard practice).
Don't worry, you cannot clip busses, as most of DAWs are 32 or even 64 bit float point, but try not to overload input of your plugins (except in some cases, when a plugin is modeled so that it reacts in a beneficial way when you overload its input).
 
Hey Chris, I've been on your forum, I dig it. Welcome!

Pretty much what Uros said. I like to track pretty hot because I think it sounds better through the converters. Any recording with 24bit resolution is going to give you heaps of headroom so you should be able to steer well away from clipping and still be well above the noise floor. There's a few threads on here about gain staging that you should read.

I'm hitting the -10 to -6 range.
 
VERY helpful and insightful article
http://emusician.com/tutorials/emusic_daw_math/
gives you an understanding of how your DAW works with sound

EDIT:
if you don't feel like reading it basically states that the higher the magnitude of the audio signal the more efficiently the bit rate functions (as long as you do not overload)
the loudest you can get it without clipping the better the the bit rate will function
 
But recording hotter will mean you're not getting as clean a signal coming from your AD converters. So in essence you're capturing a more distorted signal at higher resolution.
 
right right
very true I see what your saying the hotter you record the more pre amp you will hear, but in theory if you had absolutely transparent pre amps you would want to record as hot as possible or you want to get the most coloration out of pre amps
or am I not understanding?
 
what I'm thinking in my head is that your sonic audio quality is based off of the quality of your equipment
so what the computer is recording is what its hearing come from your A/D converter so the audio signal wouldn't be as corrupted or distorted at a higher signal strength because the computer is writing exactly what it is hearing and if it isn't overloading the bit depth (which is clipping) and you are using the bit depth to its full potential then there is less distortion in the audio signal then there would be if it is soft. however! if your equipment is warming the sound (like a tube pre or something along those lines) you will get sonic distortion and not signal distortion.
what I'm trying to say is that there is distortion in the signal no matter what because theres no way 24 bits will perfectly emulate the sound perfectly but if you push the bit rate to its highest potential it get as close as it will to emulating it
and audible distortion is different than signal distortion

but again thats what that article of information conveyed to me when I read it

EDIT: and it would lower your noise floor
 
You're nearly there, not quite though. I'm not talking about saturation from preamps (good distortion, if you've got a API or Neve or something). I'm talking about distortion caused by running hot levels into your AD converters (bad distortion)

Even if you have a very transparent preamp, it still has to go through the AD converter to be sampled and turned into a digital signal.

AD converters are made up of analogue components, and anything analogue in audio is typically calibrated to give it's best performance at 0VU. Which when compared to digital levels comes to about -18 to -14dBfs RMS depending on how your converters are calibrated. (check out this chart for a comparison of different audio levels http://media.soundonsound.com/sos/oct03/images/qadbcomparison.l.jpg)
If you're recording hot as possible then you are distorting the analogue stages in your preamp and AD converter.
 
ahhh! alright sick!
that makes sense I see now that I wasn't paying attention to the fact that it obviously starts as analogue and the way analogue pieces of equipment function
 
How much headroom should I be allowing on my raw rhythm tracks before sending them to the master mix?

If your tracks are clean and not clipping I find that if I put a trim plug on the last insert you can set the output of each track at around the -18dbfs mark. This stops you blowing out the master and lets the track faders work better within the range they were intended to be used.
 
You're nearly there, not quite though. I'm not talking about saturation from preamps (good distortion, if you've got a API or NEve or something). I'm talking about distortion caused by running hot levels into your AD converters (bad distortion)

Even if you have a very transparent preamp, it still has to go through the AD converter to be sampled and turned into a digital signal.

AD converters are made up of analogue components, and anything analogue in audio is typically calibrated to give it's best performance at 0dBu. Which when compared to digital levels comes to about -18dBfs. (http://media.soundonsound.com/sos/oct03/images/qadbcomparison.l.jpg)
If you're recording hot as possible then you are distorting the analogue stages in your preamp and AD converter.

So say your peaks are at -2dBfs in your DAW, that means the analogue components are running at +16dBu! If you've ever mixed sound at a gig and had your desk outputs hitting this kind of level then you'll know that it doesn't sound pretty! For example on the Allen & Heath mixer I work on regularly the highest level on the meters is +9dBu. After that you're hitting the clip warning light!

i'm kinda puzzled bout this.. i always thought the same, track around -18dbFS, but i.e. some people advise to track your di's for reamping (yeah, talking bout the "get your tracks ready for reamping guide) around -4 or -3.. confused :err: :)
 
So say your peaks are at -2dBfs in your DAW, that means the analogue components are running at +16dBu!

You are mixing up peak and RMS levels. Converters are typically calibrated −18 dBFS=0 VU=+4 dBu. But, we are no talking about peak levels, but RMS. When something has level of -18dBFS RMS, it can peak at -6dBFS or somewhere else, it depends.

@Cuban Dude:
Have your tracks peak anywhere from -12 and -6 dBFS, and you are perfectly safe. If you must, have your peaks go all the way down to the -3, but try to make it anywhere between -12 and -6 dBFS. I am talking peak level, not RMS now.
 
^ yeah, was talking bout peak level too, just wanna hear somebody that i'm not completely off track... thanks for the advice.. i guess i'll shoot for -6 dbFS next time with my di trackings, just feel like you get "more" of the sound when reamping if your tracks don't peak too low
 
King Diamond > *.

This thread is relevant to my interests, because I thought as long as my DAW didn't show my input signals going over 0.0dB (Axe-Fx analog -> Analog inputs) I was good. :err:

If one uses the 24-bit/48kHz digital outs to an interface, can one track to 0.0 dB, in other words leave no headroom on a track?
 
Yup, I generally shoot for somewhere around -12 but as long as it's not coming that close to clipping you should be good to go.
It wasn't till I started using outboard compressors that I really got the level differences. If you track hot into the daw when you go to use the outboard you start getting gain reduction really early- well above a threshold of 0.
 
Edited ramblings about analogue audio levels

When things are hitting your analogue stages hard there is going to be some loss in sound quality, unless you have VERY good preamp's and converters that are designed with lots of built in headroom well above +4dBu/0VU RMS (their optimum operating level, where they are designed to give good signal to noise ratio and low distortion) and even if they are designed with such good headroom they probably still wont sound their best with levels that high.

Unless you're talking about driving a nice character pre into saturation, which is a different kettle of fish. But even when people drive their fancy preamp's hard they are using in line pads after the preamp in order to NOT SMASH THE SHIT OUT OF THEIR CONVERTERS!

What makes it more frustrating is that instruction manuals for digital recorders and DAWs often tell users to record as hot as possible without clipping, as this is much easier and less confusing than educating people about proper gain staging and the the relationship between analogue and digital audio levels.

I remember first encountering this and having a bit of a wtf moment a few years ago, when using my analogue mixer as external preamps for my pro tools setup, and wondering why the pfl on my desk was giving me a clip warning light but I wasn't getting maximum level in pro tools? The problem is that nowadays we're using the clip indicators in our DAW and often don't have any at all on our pre's, which is where we should have them! The concept of headroom has been pretty much lost with a lot of people who are into recording that have never used much or any analogue equipment.

In the digital world headroom is simple. It's perfect, totally clean, until you clip. At this point you've lost information. Your signal is now fucked, somewhere in there is a square wave. Shit.
In the analogue world it's not the same. As the level gets higher and higher above 0VU your signal will become progressively more distorted, the frequency response can change, and things just generally start sounding pretty dirty, before you get to the point where the wave actually becomes audibly clipped. So long before you're hitting the digital maximum of 0dBfs you're already losing sound quality in your analogue stages. Even though your signal is fine according to your DAW.


There's no real need to ever be running signals this hot. I can't help but feel there are some guys on here who don't seem to grasp quite how much headroom we have available in a modern recording system.

Recording hot is a unfortunate habit, left over from the early days of digital recording when recorders were only 16 bit and the noise floor available was much poorer than we have now. And as a practice it's really not necessary anymore.

Digital audio ramblings

In 24 bit audio we have 16,777,216 amplitude steps available, (yes, nearly 17 million!) giving 144dB dynamic range (greater than that of the human ear, which is only around 130dB) resulting in 116,508 amplitude steps PER DECIBEL. So given a signal that is using all of the headroom available, with one single sample hitting our digital maximum, our sampled signal is accurate down to 1 one hundred and sixteen thousandth of a dB.

Compare this to 16 bit which is only 65,536 steps and 96dB dynamic range. So we have more amplitude steps available PER DECIBEL than people used to have in their entire word length for a digital recording system. We have nearly twice the resolution and detail available PER DECIBEL than a 16bit cd.

Each bit in the word length gives us 6dB better signal to noise ratio. So if you record with your absolute highest peaks at -12dBfs, you can capture a dynamic range of 132dB.
This is nearly double the typical dynamic range of an orchestra (which can be as much as 70dB) And we still have 31,775 amplitude steps per decibel here. Around half of what old 16 bit recorders had for their entire dynamic range!

Also bear in mind we're NOT recording an orchestra here. We don't have lots of low level detail required. It's noisy rock and metal that's going to be compressed and limited within an inch of it's life, dithered down to a 16 bit cd if you're lucky and ripped to a low quality mp3 if you're not.

Hell Chris Lord Alge dumps all his stuff to 16 bit digital tape before mixing. And I don't think anyone can say his sound suffers from not "using all the bits"

Another great advantage of tracking with lower levels is that come mix time you won't have to turn everything down to avoid excessive levels in your aux and master tracks.


Capture things cleanly and stop being a slave to the bits!
After all if you're trying to use as many bits as you can (because digital audio is inferior to analogue right?) then what's the point if you're completely ignoring the signal at the analogue stage in the first place! Get the levels right there and your DAW will capture it cleanly and effectively, even if you're not pegging the top of your meters. There's no point in capturing the highest resolution you can if your signal is compromised before it even hits your DAW!

It's 2011, it's 24 bit, back off your levels and enjoy the headroom! :kickass:
 
But saying "it's only distorting on the peaks" isn't really a good mantra is it? It shouldn't be distorting at all!
I am puzzled with this statement :err:
If it doesn't distort on peaks, it also means that it doesn't distort on RMS levels. No audio can have its RMS level higher than its peak level.

...
So if you record with your absolute highest peaks at -12dBfs, (which is still +6dBu in your analogue components by the way, so still hotter than optimum) you can capture a dynamic range of 132dB.

I agree with the rest of your post (except that the most gear is calibrated to operate at +4dBU=0VU, not 0dBU; you are mixing up VU and dBU), it's just that again, this ^^ is not true. Average, not peak level of -12dBFS is equal to 6 VU=10dBU. Watch your RMS meter to find optimum level, not the DAW meter (Sonalksis FreeG, Voxengo SPAN, etc.)
That said, have your tracks peak at -12 or similar, and you are perfectly fine.

I am aware that analogue 0 is not equal to digital 0 in any case.

As far as digital world goes, yeah, in the world of 24bit audio you would have to track reaaaly low to make noise floor fuck up your recordings, so tracking hot is totally unnecessary.

I try and record in [-12,-6]dBFS range because yeah, it's healthy for your converters, and it's just healthy practice to do all around (especially if you work somewhere on an all analogue equipment in the future, this practice 'translates' perfectly), so it doesn't hurt.

Plus, I try to have my channel faders at 0 before I start mixing. I chase all my start levels on gain trim pots (and that's exactly how would you do it on an analogue desk), and only use faders for latter balancing. This is also because DAW faders have best resolution at or very close to 0, which is important for automation.