A little something about gain-staging, for those who don't know

[AudioPhile777]

I don't get any of this too, I feel kind of stupid
Sure, clipping tracks are always bad, but I don't get why a track (in the digital domain) should sound better if you just lower its volume. I mean, I don't care why it should sound better, I just want to know how I should apply the things that have been said in the article / this thread

If you don't care to know 'why', you will never really know 'how'.

 

[The Arrow]

well, I'm strongly against tracking at -18 dBF because of digital bit depth. when I track at say so -5 dB I get more clear dynamic range representation, so the digital sound works i guess. anyway trimming hot tracked file is a good idea, the only question is - how does the DAW works. For example Cubase's master bus is 32 bit float and I track in 24 bit, so Cubase caclculate the overal mix once again, before it gets to master bus. So If the overall mix level is about -10 dBFu average wouldn't I loose the dynamics representation I got on tracking?
PS sorry for bad English

 

[slashvanyoung]

well, I'm strongly against tracking at -18 dBF because of digital bit depth. when I track at say so -5 dB I get more clear dynamic range representation, so the digital sound works i guess. anyway trimming hot tracked file is a good idea, the only question is - how does the DAW works. For example Cubase's master bus is 32 bit float and I track in 24 bit, so Cubase caclculate the overal mix once again, before it gets to master bus. So If the overall mix level is about -10 dBFu average wouldn't I loose the dynamics representation I got on tracking?
PS sorry for bad English

The whole "fill as many bits as possible by tracking as hot as possible" is nonsense. Especially if you track in 24 bit. At such a high resolution, the possible negative side-effects of overriding the converters are way worse. Converters are designed to work best at 0 dBVU, so unless your system is calibrated much higher than usually (which is -20 to -18 dbFS equalling 0 dBVU), you're driving the converters already quite hard if you're average peaks are as high as -5 dBFS.

So instead of making the sound bigger and clearer by embracing the whole dynamic range, you might achieve the opposite by driving the converter out of it's "comfort zone".

 

[Sam Bottner]

If you don't care to know 'why', you will never really know 'how'.

awesomeness

 

[Star Ark]

I don't get any of this too, I feel kind of stupid
Sure, clipping tracks are always bad, but I don't get why a track (in the digital domain) should sound better if you just lower its volume. I mean, I don't care why it should sound better, I just want to know how I should apply the things that have been said in the article / this thread

Just try to follow the rules set out in this thread or the articles people have supplied links for and decide if it sounds better. If you like the sound of it maybe you will feel like researching it more thoroughly. You may find that your tracks breath a little better and that when you crank the volume your mix doesn't turn into Death Magnetic

So you want to aim for -18 peaks when tracking, and you want to stop overloading your plugins (red flashing lights on output/input). Do that for a mix or two and see what you think. That's the simplest way to know whether you should do something audio/mixing related, you do it and listen to the results

 

[jval]

Good article for sure. It's a good idea to track around the level he suggests (the default PT calibration is actually -18 iirc), although I admit I often go well above that so it's a good reminder. It makes sense to keep your pre's, compressors, and converters operating at the level they were designed for.

Once we're talking about the digital world though, it would be interesting to get some more proof on the claim that plugs are designed to work at this level. I've only read this in the manual of one plug-in - Waves V series. Would be cool to get a plug-in developer's take on this.

 

[Ermz]

I really wonder whether this is the case with some older Waves plug-ins. At low levels the L1 will barely meter, and similarly with the CLA Comps series. The noise floor on those is huge if you mix down at -18. Sure, you can turn it off, but it makes you wonder what they were calibrating for.

 

[jval]

yeah, even with the SSL plugs too. In some cases I have to trim the input UP in order to get the amount of compression I want.... with some plugs it's hard to believe they're calibrated to expect a -18db input.

 

[Ermz]

Yeah I need to do that all the time with the SSL Channel plug. So that's the discrepancy here. We're dealing with a bunch of people who can't decide on a universal standard for calibration in digital levels. For instance my RME converters don't even go as low as -18dBFS = 0VU. The best they do is -15dBFS = 0VU, and that's on the obscure 'Lo Gain' and 'Hi Gain' settings. RME adamantly recommend using the calibration that amounts to -9dBFS = 0VU, so it makes you wonder.

 

[gabriel g.]

Yeah I need to do that all the time with the SSL Channel plug. So that's the discrepancy here. We're dealing with a bunch of people who can't decide on a universal standard for calibration in digital levels. For instance my RME converters don't even go as low as -18dBFS = 0VU. The best they do is -15dBFS = 0VU, and that's on the obscure 'Lo Gain' and 'Hi Gain' settings. RME adamantly recommend using the calibration that amounts to -9dBFS = 0VU, so it makes you wonder.

this is because the industry standard for ad conversation is +4dBU . So -14dBfs equals +4dBu

and just a reminder for tracking.
the best micpres have a SNR of 110dB.(and this is theoretical, because in most situations your room has a noisefloor around 10-40dB)
24bit gives you a dynamic range of 144dB

So tracking hot is nonsense. pushing the preamps is another story...

as for the plugins:
I know what you guys mean. with the simulated noisefloor of the analog plugins it is sometimes difficuilt to get the input level right because most of the time the noise floor is static.

I noticed if you tracking technical right this problem is not that big...

 

[tempe]

In regards to ideal level whilst tracking, I think its best to just aim for your converters optimum level it states in the manual, as the article said most of the time that's -18dBFS, but like Ermz I use RME converters and they to recommend tracking at around -9dBFS.

Another question, albeit slightly off topic. I've read that when recording DI tracks that are to be reamped you want your peaks at around -3dBFS, is that a load of shit or just something that's specific to preparing for reamping?

 

[Ermz]

this is because the industry standard for ad conversation is +4dBU . So -14dBfs equals +4dBu

Don't think I'm getting you there.

+4dBu is the de facto standard line level for professional analogue equipment. It does not correlate to anything in dBFS apart from what the converter manufacturer decides it should. There is no standard in digital for optimum levels, nor any straight conversion formula to transpose analogue levels into digital levels.

Bob Katz tried to popularize the K-system, but it seems to have only taken off in niche circles. For me, I don't like the fact that it scales depending on what's being mixed. People usually get levels wrong even with a single standard... imagine them trying to balance 3.

This is why this whole thing is a mess. Nobody can truly decide what to do, so the closest we've been able to get is by approximating the headroom of high quality analogue mixing desks, which tends to roughly be in the ~18dB range. Hence line level at +4dBu = 0 VU = 18dBFS. But obviously not all the plug-ins recognize this, nor do all the converter manufacturers.

 

[Ermz]

Found this relevant article written 10 years ago: http://www.soundonsound.com/sos/may00/articles/digital.htm

Apparently we already do have standards for digital, but if so, barely anyone knows about them even now.

As on analogue systems, it makes sense to build in some form of operational headroom to cater for the odd loud peak. This, however, is where all the confusion and problems occur. Since analogue equipment typically provides 18dB or more of headroom, it seems sensible to configure digital systems in the same way. After a little trial and error the Americans adopted a standard of setting the nominal analogue level (+4dBu) to equate with -16dBFS in the digital system, thereby accommodating peaks of up to +20dBu (ie. 0dBu equals -20dBFS). In Europe we have standardised on 0dBu equating to -18dBFS, thereby tolerating peaks of up to +18dBu.

He also brings up an interesting issue that worries me.

With our converters calibrated to such low levels as being 'optimum', when we play back a mastered CD through the D/A side, we are hammering our monitor controllers, mixing desks, monitors etc. with some really damn loud analogue levels. There is no middle ground. If you're calibrated for one, it's shit for the other. Now you might think 'oh well, why don't I just calibrate my A/D to be -18dBFs and my D/A to be -9dBFS', well the problem like that is if you mix with outboard gear like me. You'd essentially get mismatched in and out levels. Can't win.

 

[drew_drummer]

I think earlier on I was just being a complete and utter bellend, and completely misunderstanding the points raised. So let me try and see if I've got this:

Hit the inputs at -18dbFS, which equates to approximately 0dbVU. Leave all faders at zero. The idea being that any mix headroom you've now got, comes from the lowered pre-amp signals.

Then each plugin you add, make sure the input and output are also -18dbFS. So effectively... the final signal hitting your DAW meter never really changes. And the fader always remains at the zero point, until it comes to mixing... where-by during the mix you'll probably only adjust it by a few DB.

Have I got this right?

 

[gabriel g.]

Don't think I'm getting you there.

+4dBu is the de facto standard line level for professional analogue equipment. It does not correlate to anything in dBFS apart from what the converter manufacturer decides it should. There is no standard in digital for optimum levels, nor any straight conversion formula to transpose analogue levels into digital levels.

Bob Katz tried to popularize the K-system, but it seems to have only taken off in niche circles. For me, I don't like the fact that it scales depending on what's being mixed. People usually get levels wrong even with a single standard... imagine them trying to balance 3.

This is why this whole thing is a mess. Nobody can truly decide what to do, so the closest we've been able to get is by approximating the headroom of high quality analogue mixing desks, which tends to roughly be in the ~18dB range. Hence line level at +4dBu = 0 VU = 18dBFS. But obviously not all the plug-ins recognize this, nor do all the converter manufacturers.

damn I was at work and only read on the fly...
what I wrote is bullshit and you are right.

 

[The Arrow]

well, I just tried tracking my guitar at -18 and there are some problems. First of all I just cant get such a low level from my guitar, I have to use -20 dB attenuator to get them and second even when i get them it's impossible to play, because none of the ampsim gives any acceptable gain at such input level. What am I doing wrong?

 

[AudioGeekZine]

Amp sims seem to be where you have to ignore all theory and tradition and just do whatever it takes even if it means clipping to get enough gain. Amp sims also seem to have way too much output gain on them as well.

 

[Pedro Teixeira]

Amp sims seem to be where you have to ignore all theory and tradition and just do whatever it takes even if it means clipping to get enough gain. Amp sims also seem to have way too much output gain on them as well.

Yep, one of the reasons I hate amp sims, they have a VERY hot output level, it's just horrible

 

[jval]

another problem is the metering in DAWs... at least in PT, -18db is way down under the yellow, less than halfway up the meter. You'd think they would scale the meters differently if those gain levels were ideal in the digital mixer... the Digi manual even suggests calibrating the I/O's at -18... if I was tracking everything that low, I'd have to have everything vertically zoomed to edit the waveforms. the display just really doesn't fit the levels they suggest.

 

[C_F_H_13]

There is a producer ( I don't remember who, but he produced a ton of stuff), that the first thing he does when he mixes is insert ''trim'' on every track of his session, and then mix using it. Seems logical now.

I can totally relate to this. I've been mixing on an SSL for a while now, and every project I get that's the first thing I do. TRIM plugins to -16 (our I/O is calibrated to -16 instead of the norm -18) on all tracks. That way I don't clip the channel if I boost more then 3 db of low end. I can turn the knob full blast and still have plenty of headroom. Plus I havn't had to do an offline level trim in forever....

another problem is the metering in DAWs... at least in PT, -18db is way down under the yellow, less than halfway up the meter. You'd think they would scale the meters differently if those gain levels were ideal in the digital mixer... the Digi manual even suggests calibrating the I/O's at -18... if I was tracking everything that low, I'd have to have everything vertically zoomed to edit the waveforms. the display just really doesn't fit the levels they suggest.

Well what digi is talking about is calibrating their I/O, no specifically recording that low. My suggestion is to take all of this with a grain of salt....

Record at decent levels, and you'll have no problems if you decide to move faders.