i thought it is best to record around -18dbfs

djclueveli

New member
i got this article from emusician.com

Compressing the Singer
A singer’s performance is unarguably the most critical aspect of a vocal recording. Crucial, too, is the quality of components in the recording chain. But another area that’s important, and that will have an impact on the editing and mixing processes is the quality of the recording itself. Was it recorded with enough level? Were the dynamics kept under control? Was there any distortion?

I asked the six producer-engineers (Steve Addabbo, Bob Power, Johnny “Juice” Rosado, Rail Jon Rogut, Rick Sheppard, and Ed Stasium) how important was it to try to record vocals as close to 0 dB as possible? The general consensus was that—given a 24-bit recording with its increased headroom over 16-bit—it was better to aim a little under 0 dB to help minimize the chance of going over, than to try to get the hottest possible level to disk.

“The old-school guys always say, ‘Go for the hottest level; use up all your bits,’ and it’s 6 dB per bit,” says Rogut. “My attitude is, if you’ve got 6 dB to play with, keep your level at least -6 dB or -5 dB.”

Addabbo echoes that sentiment: “The last thing you want to do is to push it too hard. I think in digital, we’re always fighting to get enough level on, so that we’re in a really high range of definition. So the converters can perform their optimum. But you don’t want to go over that limit, because it’s just a brick wall up there.”

Rosado says he’s too often given vocal tracks to mix that are clipped, and he routinely tells clients to record their vocals with the peaks around –6 dB. “I prefer to leave some headroom on it,” he says. “Because if it’s too hot, it doesn’t give you room to play with.”

However, too low isn’t good either, because you can lose quality. “ A lot of people mistakenly think, ‘Oh, it’s digital, it doesn’t matter,’“ says Power.


There was also wide agreement that compressing lightly to disk is a good idea to tame a singer’s dynamics and maintain optimum levels. “It’s important that people understand that the issues that surround good consistent level on the input of a digital converter,” says Power. “Obviously, whatever you do inside the box is way, way, downstream of that.”

Addabbo and Stasium both said that they supplement the compressor by riding the mic preamp during recording. “I’ll always ride the vocal,” Stasium says. “I’ll push it up a wee bit, and then during loud passages, I’ll back it off. I want to get that level so that it sounds like a naturally flowing vocal.”

There was general agreement that the compression settings for recording vocals should typically have ratios between 2:1 and 4:1. Make sure to set your compressor carefully, because you can’t undo compression once it’s recorded onto a track.

“There are two ways to really mess up compression on the way in,” says Power. “One is to slap something too hard most of the time. In that case, you don’t hear any pumping, but the other negative artifacts of compression, particularly harshness, sibilance, and lack of dynamics are accentuated. The other is to try to set the compressor too subtly, without compressing too much. In that case you can end up with pumping, which can result from a combination of factors of attack and release. It’s a tricky thing.”—Mike Levine

they say to record vocals kind of hot so u wont loose any quality. they suggested to keep the level around -5dbfs but everybody on here always says to keep the level around -18dbfs so isnt that too low?
 
As I understand it, given resonable levels at 24 bit, on the digital side there is no loss (or gain to be had to run them hot). However there can be issues downstream with tracks that live well above nominal (-18 or so average). On the analog side, above nominal' is (or can be) pushing it.
The idea is to run your analog where it works best, the conversion takes care of that end, and the tracks come in where they should in the software -no need to do a bunch of trims to get them back down where they belong.
 
Last edited:
i try not to worry about it much......but i do worry about in my analog mixing desk.....which is where all my mic-pres and all my inputs go. So i always set the recording track channel (on the mixing console) so the input meters on the track is pumping up near 0 db, but never peaks over it.

once that info is good to go, ill open up my DAW (Sonar 6 PE), and have a look at the reading in there...........and the track shows it reading at about -8 to -6 db (full scale), and it usually never reaches past -6 db in my DAW.......

so it works for me, no digital overs in my DAW, and its reading nice-n-hot on my mix console .........and thats all that matters to me...
 
Except for the compression thing, those statements seem reasonable. Note that even with 24 bit ADC's the signal to noise ratio is significantly less than the theoretical 144 dB (unless you really have a high end converter), thus wasting too much headroom isn't a good idea, ie. having a peak somewhere between 0 and -6 dB is usually fine.

Compression really isn't necessary as the mic pre amp probably has even more noise than the ADC anyways. You still can compress during mixing if you feel it helps the sound, and undo it in case it doesn't work out as desired.
 
I wouldn't worry about it too much. Most people have plenty of other shit to worry about before the difference between even -24 and -6 really affects the overall sound......Much bigger fish to fry. :D
 
what exactly does mixing hot mean. i mix hip hop and i know the kicks have to be pretty loud so that means i have to raise the other sounds but what is actually considered to be hot? where should the highest instrument in the mix peak at and should any track ever go into yellow? i use fl studio and hypersonic so is there any such thing as headroom if i do not actually record my sounds but instead i just plugin hypersonic vst and insert the notes on the piano roll of fl studio?
 
I think it's (almost only) the "old school" guys** who say *DON'T* record hot and "use all your bits."

Work at the level the *gear* wants to work at - Which, at line level (depending on how the converters are calibrated), is going to give you a signal of around -18dBFS(RMS).



** Okay, maybe "old school" DIGITAL guys want to "use all the bits" -- I would think (at least, I certainly hope) that those baptized in analog know that to not be the case...
 
If the style calls for 'in you face' then the mix needs to go there. You control the density (style, tone, comp/limiting etc.) at the tracks, at the miix. ITB you can have this happen with headroom below 0' then move it up as the final move.
 
what exactly does mixing hot mean.
A hot mix is a loud mix. The counterpart is a soft mix aka. a quite mix. The latter is usually prefered since there is more room for dynamics. See my signature for a demo.

Recording and mixing are different things, though. Even if you are going to do a hot mix which I wouldn't recommend, you shouldn't exactly clip your recordings.
 
I think the old advice is as good as it gets. Track as high as you can but dont clip.
Once we start giving out numbers like -18dbFS or -10dbFS some will think that it's some sort of magic figure that results in a perfect track. Give people the reason behind the practice not just the practice itself.

The truth is, a signal that peaks at 0dbFS and goes no higher is the best level, theoretically anyway, for tracking at minimum converter noise.
The problem is, in a live recording session, arranging that situation is almost impossible. Hence the safety margin.

But the size of that margin people will always disagree on. It's only ever a guesstimate as to how much higher the signal might go, and that varies greatly depending on so many factors.

Maybe we need a crystal ball plug in which seaches for the peak level before the talent has even stepped up to the mic. Now that would be some tricky software.

Tim
 
they say to record vocals kind of hot so u wont loose any quality. they suggested to keep the level around -5dbfs but everybody on here always says to keep the level around -18dbfs so isnt that too low?
Thy're not really saying anything all that different there, dj. When they were talking about -5 to -6dBFS, they're talking about peak level.

When someone is talking about averaging -18dBFS, they're talking about the average level (dBFS RMS), not the peak levels.

We've talked about this before, dj. Remember, it's all about the converter and how it's calibrated. If it's calibrated so that an analog signal at 0VU/+4dBu converts to a digital level of -18dBFS (which is common, but not universal), then that's how it should convert, and everybody is happy.

That 0VU --> 18dBFS conversion is designed like that on purpose; so that you DO have plenty of room for peaks. If you're coming in at -18dBFS RMS and your signal has a crest factor (the difference between the average level and the peak level) of a full 12dB - which is a whole two bits of your binary sample, BTW - that would put your peaks at -6dBFS, which is exactly what those fader jockeys are saying.
Tim Gillett said:
The truth is, a signal that peaks at 0dbFS and goes no higher is the best level, theoretically anyway, for tracking at minimum converter noise.
Not necessarily true at all. The best theoretical level is to push an average analog signal voltage into the converter at the line level for which it's designed. Typically this will be +4dBu for commercial and prosumer gear, -10dBV for Soundblaster-class sound cards.

Then it's up to the calibration level in the converter to decide what that analog line level voltage will convert to in dBFS on the digital side. Typically they will convert a 0VU line level to somewhere between -15dBFS (DAT and ADAT) to -20dBFS (many newer hardware DAWs and digital mixers), with the SMPTE recommendation of -18dBFS being, if not universal, at least fairly common and a good average.

With that in mind, if one keeps the output of the converter and the input of their recorder or software at unity gain, that means one is recording at an ideal level based upon the RMS conversion level, with peak headroom equivalent to the absolute value of the digital conversion level.

But, you might say, I don't need 18dB of headroom because my crest factor is only 12dB, so I'm wasting 6dB. The problem there is, if you boost the digital input gain by 6dB, all you're effectively doing is increasing the noise floor by 6dB. If you instead boost on the input to the converter instead, that means you're sending an analog signal averaging +6VU into the converter, which, depending upon the quality of the converter and it's own internal gain structure, can easily be too hot to handle without comprimise in the way of noise or headroom.

Botton line, it's usually best (on paper, anyway; it really depends upon the gear and the signal content) to just run the converter at the average analog level it's designed for* (around 0VU) and then leave the downstream digital levels at unity. The exception there being if you're recording something so dynamic that the 18dB (or whatever it may be) headroom is still not enough. Then you can decrease the gain to make sure everything fits under 0dBFS.

*I know, I know, some folks like clipping their converters as a weapon in The Volume Wars. But that's a different topic from getting clean conversion and recording levels.

G.
 
Last edited:
The truth is, a signal that peaks at 0dbFS and goes no higher is the best level, theoretically anyway, for tracking at minimum converter noise.
I'll put this a question as it's too easy to come off sounding like a know it all.. :D
Everything I've gathered says that raising the conversion error noise 6 (or 10 or ? db) buy tracking 6 (or whatever) lower is irrelevant if the the conversion noise is buried some 20 or 30 db below your analog floor.
The implication is that this conversion noise is a gnat in a sand storm.
 
Tim, I would consider the advice you gave to actually be bad and misinformed advice. It seems like you are thinking in only one dimension with wanting to track as close to 0 as possible. You are only looking at things rom the converter's perspective. In that case I would agree with you. However, there are FAR more important things to consider, namely the analog front end. In order to push a signal that high for the converters sake you will be far exceeding the analog front ends sweet spot. Basically, an analog unity level "generally" equates to "about" -18dbfs. This means that to hit just below 0 on a converter you would have to push the front end to +18 which is WAY above a preamps optimal range. Preamps will go higher than that, but in most all affordable ones especially, they radically change in sound at leels quite a bit below +18. Now a converter on the other hand, at least current average 24 bit converters, will be much more uniform at say -18 than a comparable preamp at +18. Basically, you lose a very small amount of resoltion by lowering a converters signal by 18 db but you lose a ton of quality by overdriving the analog front end by this amount.

Djcluevelli, in my opinion the article you are reading is actually confirming more what you have been taught here than opposing it. My guess, like Glen said, is that the article refers to -6 db as a peak value whereas when many people here refer to -18 as a tracking level that is intended to be an average, and not a peak value.
 
Djcluevelli, in my opinion the article you are reading is actually confirming more what you have been taught here than opposing it. My guess, like Glen said, is that the article refers to -6 db as a peak value whereas when many people here refer to -18 as a tracking level that is intended to be an average, and not a peak value.
I more than agree, I definitely read it to be specifically referring to peak levels:
The general consensus was that...it was better to aim a little under 0 dB to help minimize the chance of going over, than to try to get the hottest possible level to disk.
The idea of shooting a bit short to avoid overs is definitely referring to peak signals.
“The old-school guys always say, ‘Go for the hottest level; use up all your bits,’ and it’s 6 dB per bit,” says Rogut. “My attitude is, if you’ve got 6 dB to play with, keep your level at least -6 dB or -5 dB.”
Once again, the issue is whether to push the peaks to use up all the bits or not. This guy says pull the peaks back 5 or 6 dB. He's definitely talking peak levels.
Rosado...routinely tells clients to record their vocals with the peaks around –6 dB.
'Nuff said.

So, yeah, a peak level of -6dBFS will typically mean a 0VU-to-dBFS conversion factor of somewhere in the mid-to-late teens. They are just looking at it from the peak perspective rather than the conversion perspective.

That works fine as a "close enough" method for those with converters where the conversion calibration is not specified or implied by other specification (e.g. maximum output level, etc.), or when the calibration cannot be tested via a 1kHz sine wave test tone. If you figure you're peaking somehwere in the -3dBFS to -9dBFS range (with your digital inputs set at unity), that you're probably driving the analog input of your converter at somehwere within a few dBs or so of what it's designed for.

But if one has the actual spec available, or can run the tone calibration test, then IMHO it's best to look at it from the perspective of the converter calibration, and not from the perspective of peak levels. Why? Because there is no guarantee of what kind of crest factor the incoming signal will have, and therefore there is no guarantee that a peak level of (for example) -6dBFS means that you are driving the converter at a nominal level actually near 0VU.

For example, recording a signal that has moderate analog compression applied to it can save several dBs off the peaks. If you have 12dB of natural dynamics and you compress the peaks by 6 dB, that means that setting your peaks to -6dBFS means that your averaging a level of -12dBFS. That means on a converter calibrated to -18dBFS, you're actually pushing it at the input at +6VU, which can be a bit hot for some converters, and maybe more importantly is boosting the upstream analog noise floor by 6dB. If you have one of the new breed of DAWs whose converters are actually calibrated to -20 or -22dBFS, then you're actually pushing the inputs to +8 or +10VU.

And it works in reverse on high dynamic signals. A signal with a crest factor of 18dB going into a converter calibrated for -18dBFS would automatically want to peak at 0dBFS on the digital side. While one may want to throttle it back a dB or two just to be safe, pushing the peaks down to -6dBFS is unnecessarily underfeeding the converter input in such a case.

G.
 
Last edited:
The guy who started this thread was talking about the digital peak levels and nothing else. The discussion that he quoted also only talked about that.

True, there is the issue of the analog gear that runs ahead of the A/D converter and IMHO it's just as important an issue as the converter peak levels. But it's not the topic that was raised!

Shouldnt we start a separate thread discussing the matching of analog pre's etc to converters? Are you guys taking me down because unlike you I stuck to the topic?

Tim
 
Back
Top