Recording Levels in DAW's

themdla

New member
Hi guys,

I've been recording for a few years and have never really gotten the results i was looking for. Recently i became aware that i may be recording and mixing too "hot". i've always tried to get the signal coming in as hot as possible without cliping, i was always told this was the method for getting the best s/n ratio.

however i have never really understood "headroom" and "dynamics" as they relate to recording and mixing. recently i read an article that suggests peak signal level should reach about -15dbfs during recording to maintain headroom.

so what do you guys think is optimal signal level coming in?
how much headroom should i maintain during the process?

Gear: Digi 002, KRK RP5, PC, ProTools LE 8

thanks :D
 
Assuming you're recording at 24 bit, which you should be, it's awfully damn hard to record things too quiet; when in doubt, turn it down, but there's almost never a need to turn it up.

That said though, IMHO, the optimum level for recording is to record at an RMS level equal to the calibrated conversion level of your A/D converter. This is typically somewhere around -18dBFS, +/- 4dB, depending upon the age and source of your converters.

Don't worry about the peak level when you set your levels this way; as long as they don't clip, you're fine. But if you're getting clipping or close to clipping, then turn the recording level down and you'll be just fine.

G.
 
Most of the "record as hot as possible without clipping" stuff came from reel to reel tracking scenarios and, to a degree, 16 bit digital. (if I'm remembering right)

"headroom" is the room you leave at the top of your scale...i.e. from -15 up to say -1 (ish)
You'll want the headroom for when you have all your tracks summed together as a mix, applying EQ and compression, limiting etc...

"Dynamics" can be the life that you've left in the song because you didn't squash the hell out of it to make it louder. And you'll know because when you look at the wave form, it'll look like a block instead of peaks and valleys (which are the dynamincs ;))
Dynamics on my board refers to my compression, MBCX and limiting section.

:drunk:
 
:laughings:If it's to loud ....then your to dam old!!!:laughings:

Just kidding. What Glen and Dog breath have stated is the recording gospel truth

Who said that?:rolleyes:







:cool:
 
Assuming you're recording at 24 bit, which you should be, it's awfully damn hard to record things too quiet; when in doubt, turn it down, but there's almost never a need to turn it up.

That said though, IMHO, the optimum level for recording is to record at an RMS level equal to the calibrated conversion level of your A/D converter. This is typically somewhere around -18dBFS, +/- 4dB, depending upon the age and source of your converters.

Don't worry about the peak level when you set your levels this way; as long as they don't clip, you're fine. But if you're getting clipping or close to clipping, then turn the recording level down and you'll be just fine.

G.

i have recently been recording 44.1khz @ 16-bit. although i can record as high as 96khz @ 24-bit. i've always been confused about dithering and started tracking at 44.1 @ 16-bit to remove the need (i'm not sure if that's really correct though). also my drum samples/vst are easier to deal with (in terms of processing) at 44 16.

from reading your gain structure applet i assume that recording @ 24-bit will provide a lot more headroom. is that correct?

so if i can maintain an RMS level around -18dbFS, +/- 4db. anything that peaks above that could be corraled by compression, provided it doesn't squash all the dynamics i need to maintain in the final mix.
 
Most of the "record as hot as possible without clipping" stuff came from reel to reel tracking scenarios and, to a degree, 16 bit digital. (if I'm remembering right)

"headroom" is the room you leave at the top of your scale...i.e. from -15 up to say -1 (ish)
You'll want the headroom for when you have all your tracks summed together as a mix, applying EQ and compression, limiting etc...

"Dynamics" can be the life that you've left in the song because you didn't squash the hell out of it to make it louder. And you'll know because when you look at the wave form, it'll look like a block instead of peaks and valleys (which are the dynamincs ;))
Dynamics on my board refers to my compression, MBCX and limiting section.

:drunk:

yeah i've always noticed that once i start adding effects (as plug-ins in the box) and their resulting gain i start to move in to cliping unless i reduce the output of effects in the chain.

as an example i normally record a DI guitar track and during mixdown start adding "amp" effects (amplitube/guitar rig/eleven). once i add an amp effect i typically start cliping. especially if i'm adding eq or distortion (ie additional signal infomation/gain) and i reduce my output volume from the effect until it's just not clipping anymore.

however i am wondering now that i know i need to maintain headroom and dynamics....

should i try to match my effect output levels to my initial signal level RMS?

or will adding effects always add additional gain and headroom is planning for gain increases?

also, in a final mix how much headroom is normally left sans buss compression?
 
:laughings:If it's to loud ....then your to dam old!!!:laughings:

Just kidding. What Glen and Dog breath have stated is the recording gospel truth

Who said that?:rolleyes:







:cool:

lol yeah i've been in that camp for awhile :D

i've always tried to push tracks as hot as possible without clipping and then tried to turn them up as loud as possible in a mix without clipping the master.

i've always used compression but i think in the wrong way.

but then again i've always had mixes that lacked depth and sounded kinda harsh.
 
from reading your gain structure applet i assume that recording @ 24-bit will provide a lot more headroom. is that correct?
Yes it is. Think of it this way: when you're recording to digital, the total amount of dynamic range you have to work with - i.e. the size of the digital "canvas" on which you can paint your musical picture - can be roughly calculated as the number of bits times 6dB, then subtract 6dB from that result.

This means that 16bits gives you a total of (16x6)-6 dBs, or 90dB to record witin. 24bits would be (24x6)-6, or 138dB of total range to record within. That means that 24bits actually gives you an extra 48dB to work with. Or to put it another way, you can push your signal down lower in 24bit without having to worry about hitting the floor, in turn giving yourself just that much more headroom to play with without haing to worry about clipping.

This is what I mean when I say it's damn hard to record "too low" in 24bit. You have a lot of room to push the signal down without having to worry about bottoming out, giving you just that much more headroom to play with in turn.

But the thing to remember though is that RMS level does not mean peak level. When I say RMS at -18dBFS (for example), that means the average overall signal level. the peaks will easily fly above that Which brings us to...
so if i can maintain an RMS level around -18dbFS, +/- 4db. anything that peaks above that could be corraled by compression, provided it doesn't squash all the dynamics i need to maintain in the final mix.
While it is possible that you may approach clipping when you RMS at -18dBFS, it's pretty rare to find signals that are quite that dynamic. 8 times out of 10 you should be safe. However, if your input signal *is* that dynamic (technically speaking, it has a crest factor of 18dB or greater), I'd personally just turn the record level down instead of compressing or limiting the peaks.

The difference is that compression will change the nature of the signal, and therefore the actual sound. If it's a change you want, ten it's fine, of course. But f you'd rather just record the clean signal and wait until you mix until you make compression decisions (which is what I personally prefer), the just turning the record level down is the way to go. It will not change the sound at all, it will simply only record at a safer record level.

An exception to that that happens a lot for me is when you get a vocalist who is inexperienced in studio technique and tends to wander all over the place in relation to the microphone, unintentionally causing their levels to also wander all over the place. It can be really hard to keep sensible recording levels that way, and often limiting or compressing on the way in is needed just to keep things under control. But the catch there is, depending upon your DAW, your limiting plug-ins may or may not work while recording. You'd have to research that for your DAW. I avoid that by using an outboard analog compressor or limiter before the signal gets to my DAW, but if you don;t have that ger available that may not be an option for you.

G.
 
Yes it is. Think of it this way: when you're recording to digital, the total amount of dynamic range you have to work with - i.e. the size of the digital "canvas" on which you can paint your musical picture - can be roughly calculated as the number of bits times 6dB, then subtract 6dB from that result.

This means that 16bits gives you a total of (16x6)-6 dBs, or 90dB to record witin. 24bits would be (24x6)-6, or 138dB of total range to record within. That means that 24bits actually gives you an extra 48dB to work with. Or to put it another way, you can push your signal down lower in 24bit without having to worry about hitting the floor, in turn giving yourself just that much more headroom to play with without haing to worry about clipping.

This is what I mean when I say it's damn hard to record "too low" in 24bit. You have a lot of room to push the signal down without having to worry about bottoming out, giving you just that much more headroom to play with in turn.

But the thing to remember though is that RMS level does not mean peak level. When I say RMS at -18dBFS (for example), that means the average overall signal level. the peaks will easily fly above that Which brings us to...While it is possible that you may approach clipping when you RMS at -18dBFS, it's pretty rare to find signals that are quite that dynamic. 8 times out of 10 you should be safe. However, if your input signal *is* that dynamic (technically speaking, it has a crest factor of 18dB or greater), I'd personally just turn the record level down instead of compressing or limiting the peaks.

The difference is that compression will change the nature of the signal, and therefore the actual sound. If it's a change you want, ten it's fine, of course. But f you'd rather just record the clean signal and wait until you mix until you make compression decisions (which is what I personally prefer), the just turning the record level down is the way to go. It will not change the sound at all, it will simply only record at a safer record level.

An exception to that that happens a lot for me is when you get a vocalist who is inexperienced in studio technique and tends to wander all over the place in relation to the microphone, unintentionally causing their levels to also wander all over the place. It can be really hard to keep sensible recording levels that way, and often limiting or compressing on the way in is needed just to keep things under control. But the catch there is, depending upon your DAW, your limiting plug-ins may or may not work while recording. You'd have to research that for your DAW. I avoid that by using an outboard analog compressor or limiter before the signal gets to my DAW, but if you don;t have that ger available that may not be an option for you.

G.

First off thanks Glen for taking the time out to help me your advice is always clear and consice :D

Ok i understand what you mean and i'd have to agree. i doubt my signals have more than 18db in dynamic range RMS to Peak. Bad singers not withstanding lol

out of curiosity...

do you normally include compression before going to "tape" in your signal chain?

if so (material/signal/performer dependant) is it fairly light compression?

do you still maintain -18dbFS+/- post compression going in? (i'm sure this question is redundant but i'm still wrapping my head around these concepts)

BTW i do have an outboard 2-channel compressor in my rack.
 
do you normally include compression before going to "tape" in your signal chain?

if so (material/signal/performer dependant) is it fairly light compression?
I like it mostly on vocals, and mostly for the reason given, though it can come in handy sometimes on guitar when it's recorded direct, when a sloppy guitarist can send the signal levels all over the map.

The settings are indeed entirely dependent on the signal. It's hard to give even ballpark settings, because I'm not doing it at that point in the chain so much for sound shaping as I am dynamics control. But on vocals I'd say most often the threshold will be set somewhere within a couple of dBs of the estimated RMS of when the vocalist is where I want them (I only have analog VU meters to work with at that point), and a ratio usually somehwere in the 4:1 range, give or take. I just set my A&R to "Auto" for that and it's usually fine. But it really depends upon the situation. For direct guitar, I may switch to the limiter - or not - depending upon the situation.
do you still maintain -18dbFS+/- post compression going in? (i'm sure this question is redundant but i'm still wrapping my head around these concepts)
Yep. Just because I'm applying some signal control on the analog side doesn't mean I want to change my recording levels on the digital side. And I'll usually make the proper adjustment (if needed) right on the output gain of the compressor.
BTW i do have an outboard 2-channel compressor in my rack.
Very cool. Don't hesitate to experiment around with it. That's the best way to find your path. :)

G.
 
... i've always tried to get the signal coming in as hot as possible without cliping, i was always told this was the method for getting the best s/n ratio...

You probably got it already that that logic was from the tape days and doesn't matter in digital recording.

Lots of good comments on this thread. Something I'd add is that I always look at how loud I want to use the track later on. If, say, it's a shaker track that will be very light in the final mix, I might record it peaking at -30db. I try to record things so that I don't have to boost them later. Ideally I have to turn them down a bit or not at all.

This is especially true of cymbals. The softer I can get away with recording them the better they sound to my ears. It seems like the more headroom I leave on a cymbal track the truer the harmonics are. I suspect the concept of "recording everything as loud as you can" is why cymbal samples are, pretty much across the board, horrible. They record them way, way too loud, maybe triple. On a MIDI track my crash cymbals are usually at a volume of 36 to 72 out of 127 max.
 
You probably got it already that that logic was from the tape days and doesn't matter in digital recording.

Lots of good comments on this thread. Something I'd add is that I always look at how loud I want to use the track later on. If, say, it's a shaker track that will be very light in the final mix, I might record it peaking at -30db. I try to record things so that I don't have to boost them later. Ideally I have to turn them down a bit or not at all.

This is especially true of cymbals. The softer I can get away with recording them the better they sound to my ears. It seems like the more headroom I leave on a cymbal track the truer the harmonics are. I suspect the concept of "recording everything as loud as you can" is why cymbal samples are, pretty much across the board, horrible. They record them way, way too loud, maybe triple. On a MIDI track my crash cymbals are usually at a volume of 36 to 72 out of 127 max.

now the whole picture is starting to come together for me :D

see i have always done things the opposite of that and as i've stated before the results are self evident (read bad).

i have to agree with you completely. i use sampled drums exclusively and i always find that the cymbals are way too harsh and almost never sound "natural" like a big budget recording.

i've always pushed my recording signal to -6db on every thing i track to "get the best s/n ratio" lol :) and never leave any headroom. then after i add effects during mixing i have to turn the effect output levels down to compensate for the added gain without driving the track to clip.

but then again my mixes sound like crap and never sound dynamic, natural, or easy to listen to.

all this time i thought the speakers were to blame for coloring the sound too bass heavy resulting in me mixing too harsh. but really i'm starting to think its my lack of dynamics and headroom that are making these mixes sound so off.

:p:p:p
 
I've got some boring stuff on the matter here: http://www.massivemastering.com/blog/index_files/Proper_Audio_Recording_Levels.php

But in short, yes - 24 bit give you a lot more *DOWNWARD* headroom ("foot-room") and a theoretical 256x the resolution of 16-bit. Well worth the trouble. It allows recording at "normal" levels without a whisper of compromise.

dude!!!!! that article was awesome i like your writing style. :cool:

this is starting to make more sense by the second, what a fool i have been!

all this time i was using dictoms from old recording books i had read and never knew DAW was any different than analouge as far as tracking was concerned. in my mixes i literally do have to turn down every track -15db to not clip. my mixes endup harsh but dull at the same time (all mids no highs), and lacking any sense of depth or dynamics. everything fights and i have to massage (compress and eq) every track to just make it work.

wow

your article outlined my whole problem... i feel pretty stupid having spent the last few years not knowing this :(
 
i feel pretty stupid having spent the last few years not knowing this :(
Don't beat yourself up over it; the whole "use every bit/record as hot as possible" myth is not only a bad path that practically everybody takes starting out, but it's something that even still shows up in the owners manuals of some digital gear even today. Stick around here for a short while and you'll see plenty of others posting here coming up that exact same path behind you.

G.
 
I've got some boring stuff on the matter here: http://www.massivemastering.com/blog/index_files/Proper_Audio_Recording_Levels.php

But in short, yes - 24 bit give you a lot more *DOWNWARD* headroom ("foot-room") and a theoretical 256x the resolution of 16-bit. Well worth the trouble. It allows recording at "normal" levels without a whisper of compromise.

I liked the burnt steak analogy.I always overcook my steak.I'm probably guilty of burning my levels also.Good article.
 
Update!

ok so i did some tracking last night and all i can say is wow :D

this makes all the difference in the world. so easy to track when everything isn't on the verge of clipping the whole time haha.

i wish i could express how happy this information and the results have made me. many times i have wanted to give up on recording having never gotten decent results and now new life has been breathed into me. i am excited to record again! :cool:

so thanks to everyone for helping me understand this... i wish i could buy you all a round :drunk:
 
another couple questions....

now that i have drastically reduced my input levels and vst instrument levels to create headroom i've noticed that i have to crank the monitors a bit more to get a similar monitoring level as my hot recordings.

i guess this is normal but....

how do you guy compensate for this?

where do you monitor volume wise (i was monitoring last night around 75%)?

would higher wattage/larger monitors provide more volume at a lower volume setting? (the RP5's are around 45w per side)
 
Back
Top