RMS or Peak metering?

D

Drewseph!

New member
Hello gentlemen, I have a quick question. I have recently discovered that the meters in my daw are complete ass and that I should be using a metering plugin when setting gains and what have you. MY question though, is should I be setting the peaks of my tracks at -18dbfs, or should the RMS be hitting at -18dbfs?

Thanks, Drew

P.S. I am formerly known as Re-tox_stl, but for some reason that account has been locked up or something, so I am by no means a new dude :D
 
Drewseph! said:
Hello gentlemen, I have a quick question. I have recently discovered that the meters in my daw are complete ass and that I should be using a metering plugin when setting gains and what have you. MY question though, is should I be setting the peaks of my tracks at -18dbfs, or should the RMS be hitting at -18dbfs?

Thanks, Drew

P.S. I am formerly known as Re-tox_stl, but for some reason that account has been locked up or something, so I am by no means a new dude :D
Click to expand...

Hey Your Unmastered mixes should hit about 18 and mastered ones about 10-12 for home productions
 
There's no absolute correct answer but, depending on the material, having the RMS around -18dBFS and peaks in the -14 to -12 range will probably serve you okay.

Think about metering in the analogue world--you'd probably try to average around the 0dBu mark allowing some occasional peaks to go "into the red", say +4 to +6. In the dBFS scale, +6dBu is about -12.
 
Drewseph! said:
should I be setting the peaks of my tracks at -18dbfs, or should the RMS be hitting at -18dbfs?
Click to expand...

If you're asking about setting levels when recording, you should aim for peaks to be above -10. There's no reason to avoid levels right up to maximum, other than to avoid distortion if something even louder comes along.

--Ethan
 
The trick is to get the RMS around line level. The actual level on the meter will depend on how your converters are calibrated. A safe rule of thumb would be -18dbfs RMS as a target.

You will always have to make sure that the peaks do not clip (that's what the peak meters are for).

The idea is to run the analog side of the chain at line level, because that is the level it was designed to run best at. Ethan is correct in that the peak level doesn't matter (unless it clips) on the digital side, but running the preamps and other equipment on the analog side above its nominal level leads to problems.
 
Ethan Winer said:
If you're asking about setting levels when recording, you should aim for peaks to be above -10. There's no reason to avoid levels right up to maximum, other than to avoid distortion if something even louder comes along.

--Ethan
Click to expand...

Well, this is a subject for debate. You'll see a lot of people claiming that some plug ins (usually emulating analogue operations) can sound bad if levels start to approach 0dBFS.

Being honest, I've not heard this--but then I try to keep processing to a minimum and don't have a ton of extra VSTs either.

However, I try to keep my levels pragmatic. If I'm mixing a bunch of tracks, having every one of them approaching 0dBFS would mean I'd have to back off all the individual track levels anyway to avoid clipping in the additive mix or when applying effects/processing that adds any boost. It generally seems to make more sense to me to keep the levels a bit lower to start with since this has the additional advantage of giving me some protection against the occasional really big transient. However, I also don't get flustered when one of these transients gets up near 0dBFS!

Frankly, I wonder if there's an element of "fashion" in all this. Ten or fifteen years ago, everyone said "track as loud as you can to keep the noise floor low and not waste bits". Now it's "track at -18 to leave headroom for mastering and make your plug ins sound better". I don't take either as gospel and just go with the flow (taking into consideration the eventual mix and the type of material being recorded).
 
It's not fashion, it's that the equipment changed. 15 years ago, you had 16 bit converters will line level equaling -10dbfs (for the most part). With only 10db of headroom, you would be hard pressed to get your RMS levels up to line level before you clipped, so the advice was valid.

Now, with 24 bit converters calibrated closer to -18dbfs equaling line level, the whole thing has changed.

Ethan's advice is correct, if you are looking at it from a pure conversion and digital recording point of view. But if you take a signal with a low dynamic range, like a violin or a synth lead sound, and record it peaking at -1dbfs, you will be running your preamps somewhere around +15-18 dbVU (+19-22dbu), which is certainly at the limit of what the preamps can put out and 15-18db over where they were designed to comfortably work.

Having random peaks approaching 0dbfs is never a problem, but when you get the RMS level up there, that's the problem.

It's the same with the emulation plugins as well. If you run a signal will a really high RMS level into them, they will act like the hardware would if you ran a +15dbVU signal into them- they would distort.
 
No disagreement with most of what you said. If you have to run your pre amps flat out to get a signal up to nearly 0dBFS, that is a bad thing--and a totally unnecessary thing too. In the analogue domain, 0dB on a meter is equivalent to about -18dBFS--and, in analogue you record around 0dB RMS with peaks going where they need to above that, trusting in the headroom. (This being in an ideal situation--I've worked on material where there was a hard limit of about +8dBu, a constraint imposed by digital satellite limitations--but that's another story.)

The only thing that concerns me is statements that "unmastered mixes should hit about 18" (I assume he means -18dBFS). That's incredibly low for a mix--even back in analogue I never saw many mixes that didn't have peaks going above 0dBu. When you're in the mixing stage, there's no harm in eating into your headroom. Similarly, I accept that there may be plug ins out there that will clip if presented with a signal at, say, -3dBFS. However, I'd also say that they must be badly written plugins. In the digital domain, they're just performing mathematical operations on the zeroes and ones--and a plug in shouldn't care about the level until the answers in those mathematical ops try to go above all ones. Now, I repeat, I accept there may be some out there that have problems with high levels--but I don't own any in that category.

None of this is an argument for working at stupidly high levels--I don't--but sometimes the "never do it" answers are a bit proscriptive rather than showing a real understanding of what's going on.
 
Bobbsy said:
Well, this is a subject for debate. You'll see a lot of people claiming that some plug ins (usually emulating analogue operations) can sound bad if levels start to approach 0dBFS.
Click to expand...

This is debated only by people who haven't actually tested it. :D

In fairness, you said "emulating analog" so such plug-ins might soft-clip at lower levels. By for the most part digital audio is perfectly clean at all sensible levels.

Most modern DAWs use 32-bit floating point math, which accommodates levels that greatly exceed 0 dBFS. I proved this in my AES Audio Myths video, in the section starting at 53:39. Jay is correct that a system can be calibrated so poorly that the preamp or mixer clips before 0 dBFS at the converter. But you'd have to go out of your way to mismatch levels that way.

--Ethan
 
For those without much experience of analog emulation plugs here is a little test with an acoustic guitar recorded around -18dBFS RMS

This clip I used a straight non emulation plug Sonalksis Free G to push peaks way up to the top of the digital scale and then pulled the fader down to get the level back to dancing around -18dBFS RMS, everything is fine as expected
View attachment Test top headroom non emulation plug plug.mp3

This is the same clip run through an analog consol emulation plugin where simulated 0VU = -18dBFS I pushed the input all the way up to digital zero and then pulled the fader down to get back to get the level back to dancing around -18dBFS RMS. as you can hear the plugin is distorting like crazy. Apparently this is what would happen if you were to hit an old EMI consol at around +22 dBU, at least accoeding to the waves model
View attachment Test top headroom analog emulation plug.mp3

Hope this helps and watch your levels!
 
Ethan Winer said:
Jay is correct that a system can be calibrated so poorly that the preamp or mixer clips before 0 dBFS at the converter. But you'd have to go out of your way to mismatch levels that way.

--Ethan
Click to expand...
You don't have to go that far out of your way to get that to happen, depending on the sound you are recording. It will probably never happen on drums, but anything with little or no transients pushed up to 0dbfs will push the analog chain into the red.

On the analog side, the average signal level is what is important and what will generally become a problem if pushed too hard.

On the digital side, the only thing that will cause a problem is the peak level.

Plugins generally don't have a problem with levels up to and even exceeding 0dbfs, unless they are designed to emulate hardware. As part of that emulation, they will distort just like the real hardware would if you sent that level to it. Sometimes I do that on purpose to make drums crunchier or to distort a vocal, but if you don't want that effect, you need to keep your levels in the range that will keep the plugin from distorting.
 
Ethan Winer said:
Most modern DAWs use 32-bit floating point math, which accommodates levels that greatly exceed 0 dBFS. I proved this in my AES Audio Myths video, in the section starting at 53:39. Jay is correct that a system can be calibrated so poorly that the preamp or mixer clips before 0 dBFS at the converter. But you'd have to go out of your way to mismatch levels that way.

--Ethan
Click to expand...

Indeed. I used (and continue to use) one of the first DAWs that worked with 32 bit floating point (Audition, formerly Cool Edit). In this, it's entirely possible to mix to levels that vastly exceed 0dBFS then normalise downwards without lack of quality.

I'm not arguing the "track as hot as you can so you don't waste bits" rubbish. It's necessary to understand what's going on throughout the whole analogue-digital-DAW-mix-back to analogue chain to make sure you don't have unfortunate mismatches. The only thing that concerns me is that some folks are now making arguments just as foolish the other way, for example saying that no peaks should exceed -18dBFS.

The trick in all this is a sensible gain staging structure which isn't excessively high--or low--at any stage.
 
You must log in or register to reply here.
Back
Top