'Hot' Recording Levels vs. SOUND QUALITY Experiment

M

MASSIVE Mastering

Guest
REpost by REquest - Experiment concerning recording levels - Get past the "record it hot" goofiness out there and try getting a mix to truly sound the way it WANTS to sound -

-------------

The "rule of thumb" is to be at the level your gear wants to be at. Generally, 0dBVU or around -14dBFS. Soome gear is calibrated differently (some around -18dBFS and some up to -12dBFS=0dBVU) but having the "meat" of the signal around -14dBFS is a good place to be. Tracking AND mixing.

I think a lot of people don't realize how important it is to keep a good amount of headroom through the entire process. This is a fairly new developement - even in home recording. I'm really not sure where it all started, but I really hope it stops soon.

As a mastering engineer, I get projects in all the time at both extremes - Projects that have been "absued" at every possible chance - Clipped and overly compressed at the track level, mixed too hot and then limited - And then the client asks for "more volume" on top of that - which is quite difficult.

Then there are the projects that come in with airy, open sounding highs, great headroom, little unecessary compression or limiting at any stage, peaks at -6, -8, sometimes as low as -12dBFS (PEAKS we're talking about here) that have SO much potential for sheer volume that it's silly. Why? Because if you're going to use up all your headroom, it's best to do it ONCE - And that's as late in the game as possible (normally during the mastering phase).

But hey - Don't take my word for it. Do yourself a favor:

Record a project with really hot levels. Something simple (because you're going to do it twice) without too many tracks. Kill everything as much as you want. Compress tracks on the way in, during mixing, at the groups, at submixes, go nuts. Make sure you mix it through a limiter on the 2-buss.

Now record the same project using very conservative levels - Don't compress anything unless it actually *needs it* to fit into the mix. Keep the levels reasonable (nothing ever peaking above -6dBFS at the very most). Maybe put a dB or two of compression at the group or buss levels to "gel" things together (again, only if it actually benifts the mix - Not for the sake of volume) and complete the mixdown.

Put mix A (the really loud one) on an empty track in an empty project. Put mix B (the one that wasn't smashed) on another track. Turn mix A *down* to match the *apparent* volume of mix B and then MUTE it. Play the tracks, SOLO'ing mix A (so you hear it instead of mix B) back and forth.

Honestly - Mix B is going to absolutely SMOKE mix A in raw sound quality. If it doesn't, something is seriously wrong somewhere.

Now that we've established that smashing levels sounds like a$$, do something bizarre - NORMALIZE mix B (yes, it's generally a bad idea, but go ahead and do it) and turn mix A back to unity.

Now, strap a limiter onto mix B and just ram it in until it's as loud as mix A. Now we're *increasing* B instead of *decreasing* A - Just by ramming it into a brickwall limiter. Doesn't have to be anything fancy - Hell, just clip it at the track level if you need to.

Now, SOLO A/B them back and forth again and tell me that mix B doesn't hold up to the volume of mix A with MUCH better sound quality and general clarity.

The headroom you kept on the B project is like money in the bank as far as the *potential* quality and volume of the finished product is concerned.

"Some" headroom - Every track, every mix, every subgroup, everywhere - all the way through the project.

This is how it's done downtown.
 
copied over from other thread too.

please note that its slightly different when working in 16bit as the reduced resolution in short means that if you record with 14dbfs of headroom you are actually only getting a 12bit signal recorded and eating into the 12th bit as you do it.
as such you need to be much closer to 0dbfs all the time when using 16bit making the whole thing much harder to do, never mind the impact on quality which then appears from everywhere..

theres a ton of other reasons why not to use 16bit that ive covered before but yet i still see many people doing so. if you are you should not only seriously consider not doing so but also johns excellent comments dont fully apply and you need to work differently. be warned though you are making the whole job not only infinitely harder for yourself but with the pleasure of seriously impacted quality
 
neilwight said:
please note that its slightly different when working in 16bit
Click to expand...
You still want that headroom, the difference in 16bit is that the dynamic range is not 144dB as in 24bit, it's only 96dB.

Now if you don't like the resolution you're getting, then I say to you, that's why most everyone is tracking at 24bits these days.

In terms of bits - if you were to put a logrithmic dB scale next to a bit scale, the bit scale is NOT linear. For example, if you have a clipped wave (0dBFS), you cannot move that value linearly halfway down the meters and say you have only used 8bits. The bits are merely a representation of a non-binary number. (for example, 10 in decimal is 1010 in binary...5 in decimal is 0101 in binary...you've only used one less bit).

The numbers that those bits represent *are* linear, such that 0dBFS to -144dBFS (or -96dBFS in 16bit) is broken into 16,777,216 equally-wide volume values (or 65,536 in 16bit). Technically, you have set the 16th bit whenever the volume is higher than the 32,768th (2^15) volume value.

So having said all that, you have as good bit definition at any volume on the scale, it just differs how far away you are from the floor of the dynamic range. You are correct in that 16bit does have much less room to work with and 24bit when it comes to turning down to get more headroom.
 
Last edited:
i had been trying to keep the maths out :)
it usually becomes confusing or a turn off for many and i didnt want it detracting from the value of the information posted, which i hope everyone reads and follows.

im not sure i follow your post though. im also wondering if theres some confusion perhaps regarding terminologies somewhere. just to be clear, while common parlance (me included) is to express the resolution of the audio as the bit rate. this is actually correctly termed bit depth, bit rate is actually "technically" to do with the amount of data required to transmit the file per second and is usually expressed as Mbit/sec. these are very different things.

with 16 or 24 bit rates, the places are fixed as there is always a requirement to keep it relative to 0dbfs. as such with 16bit, recording down the volume scale is eating into the range available to you. while this also happens with 24bit theres no "slack" with 16bit as there is with 24bit.

im continually amazed anyone is using 16 bit thesedays and yet id wager than more than half doing work in a DAW globally are still using 16bit. i see it all the time outside pro situations and sometimes amazingly from within too. just look around this or any other forum , the bulk are working at 16 bit
it really is staggering.

that aside, if you are working with 16bit resolution, recording with peaks at -14, and then mixing up to -6dbfs, its clearly going to leave you with a noticably impacted quality compared to if you had worked close to 0dbfs, ensuring no clipping and using the full range of 16bits available. theres simply no breathing space to be cautious if you want max quality out of 16bit.
theres no luxury for leaving headroom and you at all times must be extra exact, ensuring you arent clipping, a range of -0.5dbfs is a good place to aim for with respect to peaks at final mixdown.

and all of this before you get into discussions ref noise floors, introduction of rounding errors into the audio stream that will make it onto the CD anytime you do anything within the DAW itself (ie pan, gain, reduce volume, process etc etc)

its a situation that no one should be forcing themselves into these days however. theres no need with todays computers and systems.
 
Last edited by a moderator:
You're probably right, neil, about the number of people here working at 16bit still. I should have clarified that I meant professionals. Anyone who knows what they're doing and charges for their services is (or should be) working at 24bit until the very last stage, at the master sent to the duplication house.

I try to explain the math in simple terms to start, but I keep it nonetheless. Music is mathematical; to fully utilize your gear, the math has got to be there. The ones that make good records without math either have assistants that do it for them or have such good ears that they do it without realizing it. The latter are very rare, the rest of us must resort to the math.

Sorry if I sound like a jerk; I'm not trying to be, and I really don't care what people think anyway. I shoot for perfection; I don't settle for less. In my mind, ignorance can only lead to less.
 
Last edited:
oh! it's simple:

you get nearly 6dB of dynamic-range (or, from the other side: signal-to-noise ratio) for every bit. so 16*6 = 96 or 24*6=144.

that means: take a pure 16bit signal. turn the volume down to -24dB. save it. now you have 12bit of effective resolution. even if you boost the level after the process by +24dB. of course, your file is in 16bit format. but only (a resolution of )12bit is used.

another thing is: if you add or multiply two number with different precisions, then the result has never a higher OR EQUAL resolution than the numbers involved. that mean: every mixing in the digital domain will lower the quality.

and rounding errors: say, you have a value of 30001 at a certain moment. and your compressor is actually applying 6dB of reduction to your signal. in digital, the plugin will divide your value by 2 (this is -6dB, remember a factor of two is a binary shift, you throw one bit away). but what is 30001/2? 15000 or 15001? here is a problem, and this happens often!! additionaly, most personal computers will round using a method rounding the even numbers up and the odd down (or the opposite). this is very bad for audio. simply use 24bit or more, and don't think about this ;)

computers are very inacurate machines, this is the first thing you learn studying infomatics.
 
Last edited:
thats a great post massive. very easy to understand.

question though-

MASSIVE Mastering said:


Because if you're going to use up all your headroom, it's best to do it ONCE - And that's as late in the game as possible (normally during the mastering phase).

Click to expand...

im not sure if im just lookin at it wrong or if ive just been totally confused.... haha. prolly the latter.

when you say that^, are you referring to people throwin compressors and limiters over the tracks, while they are makin em? and not to whats done after the fact, in the mastering stage?

not that i have a clue, but i was told once that if you are gonna use a maximizer or limiter or anything like that, that its better to smash it little by little. this could be totally wrong, i dont know. but does your advise apply across the board? for instance, i will often only compress the whole track a little bit at a time, at different stages of 'mastering'. not crushing it with L3 all at once at the end.... would it matter, or be any better or worse?
thanks.




peace.
 
thats exactly what he is saying, if you follow that process you will have a great track with a good clean signal, dynamics and sound. use of compressors during mixing are to control transients or shape sound, not to make your tracks bang out loud. you should leave it until the end process.

you shouldnt be putting any sort of limiter or mix buss comp onto it unless its for mixing purposes as all you are doing is limiting options available for the ME. you want to give them good dynamics and bags of headroom if 24bit

they will process and once its completed they will close up the headroom close to 0dbfs and then apply a limiter or hard clip their converters if extra levels are required.

the use of limiters in mastering are only to squeeze out extra volume at the very end if required. the main volume gains come from tightening up, balancing, adjusting, correcting issues and adding definition to the track itself through a variety of processes in the mastering chain.

usually its enough but if not then you can stick a limiter and get an extra 1-3 db out of it. anymore and you are getting into dodgy territory (whether it be with one or the total of several).

the track may sound like it can take it but you are adding a serious amount of flat tops in serious quantities and thats never good.

splitting the load across several limiters can sometimes yeild more transparent results (say looking for 3db using 1db on each limiter) but sometimes not. its a horses for courses thing.

limiting or compressing purely for level should be used as a last resort and definately at the end of the chain. sometimes even hard clipping converters can sound better if you have good ones.
 
Last edited by a moderator:
bags of headroom!! haha.

is that anything like max headroom?

anyway, thanks neil. glad i have some idea... haha.

what i do is squash em just up to the point where things are getting audibly changed, then i back it off a little. but i do that like two times throughout the process.... so i dont mud anything up too bad, but i can get a little bit of extra volume...


thanks again guys. good posts.






peace.
 
:)

sorry edited above as you were typing, hope thats made more clearer though i left headroom in there for ya ;)

sometimes, limiting until you can hear it degrade has alread done too much damage. especially if you are going to radio or vinyl for instance.

use the rule, no more than 3db in total and try for none everytime (though its seldom achievable today in some genres)
 
And don't compress "just to compress" - Altering dynamics *AT ALL* when it isn't directly benefiting the mix (just doing it for volume) is almost NEVER a good thing.

Don't squash because you *can* - Squash tracks only when (and only enough as) absolutely necessary when the mix is asking for it. If you're not sure that the mix is asking for it, just don't do it.
 
neilwight said:
sometimes even hard clipping converters can sound better if you have good ones.
Click to expand...

absolutly true. but one thing seems to go under in this discussion. i know alot good producers mixing so well, that no sum-processing is needed. and they use compression and limiting excessively, but not on the sum and wisely.

what you are saying is correct, but if you ask me, the mixing process gives you much more freedom to control the dynamics in an "invisible" and less destructive way. of course, you need the wisdom to do it right. the ideal scenario doesn't require any audio processing during mastering.

a beginner should follow your advices when releasing some tracks. but should not forget to play around with this units to learn how they work and how they affect the sound. because one day, they will be able to produce a mix with an impressive loudness and an interesting highly dynamic sound a mastering engineer will never achieve with sum-processing.

i think this is an important thing to remember. but if you are new to this, and not really sure what your doing, then ask a good mastering engineer (during mixing).
 
Last edited:
moses,
thats a very fine point.

my music background is electronic music (though i do play the cello) from the artist/production side atleast and so in this i did compress and eq excessively all the time. its part of the sound and was the only way to get it right more often than not and in mixing it breaks almost every rule going with respect to spacing, dynamics etc in a traditional sense.

ive also worked with some artists doing acoustic or minimal material in more recent times (maybe live drums and one or two guitarists) and from this perspective all the work comes in setting up mics, getting the audio to sound right going in and after that theres only been very minimal processing, certainly with respect to dynamics. most things added just ruined the sound or detracted from it certainly.

actually i once went with a friend to a large studio to work on his album with a big name producer. my friend decided he wanted to some quirky electronic thing as an intro track and this guy was suddenly so out of his depth it really was unbelievable. just with some synths and electronic sounds, nothing experimental in the least. the day before we had had a 18 piece orchestra in doing strings and he had breezed through that no sweat yet give hime some cheezy bleeps and TR drums and it wasnt coming at all.

just felt it was worth pointing out. ive found that people in music can both be very knowledeable...and correct and yet have entirely differing experiences and advice with respect to mixing, often entirely contradictory on some topics.

i guess its this that makes it a minefield for people starting out.
 
Neilwight & Massive Mastering....

I see both of you seem to be pretty knowledgable on the subject of mastering but do you think certain ME's are better at mastering certain genre's?

Even when it comes to mixing? B/c I know there are all kinds of FP headz on here from different genre's but I am wondering if you guys are giving this info out pertain more towards one genre or if this stuff is sort of universal?
 
M.E.'s can "specialize" to a point, and can be more "comfortable" with certain genres than others, but getting proper recording levels is pretty universal.

If it were analog tape, it'd be another story - In the case of digital, there is *nothing* to gain and *everything* to lose by hoarding bits.

I don't think a lot of people even realize that all other things at unity, a preamp for example, is working 12dB over nominal to hit 0dBFS. 12dB into the available headroom.

If I were that preamp, I'd be pretty pissed.
 
well, i think neil is pretty damn sexy, cuz he specializes in electronic stuff for vinyl. thats pretty sweet....




peace.
 
some nifty footwork there to keep it on topic ref recording levels. ;)

djcurve, perhaps post a new thread up about it if you like.
 
You must log in or register to reply here.
Back
Top