Limiters

[Bobbsy]

I have a slightly different take on this.

As a matter of course, I prefer to use no compression when tracking--and dynamics processing is applied after the fact "in the box".

However, I do put a limiter in circuit on many/most tracks. But...and it's a big but...I adjust my gain structure so that, in the normal course of events, nothing ever gets close to the limiter. It's purely a "belt and braces" theory to protect me on the very rare occasions somebody (usually a vocalist) suddenly lets rip and does something completely outside the levels we've adjusted for. It only happens rarely but, just occasionally, a "special" take can be saved by the limiter.

And, to answer your question, to use a compressor as a limiter, you set it up to have a fairly high ratio, no output gain with the threshold at something like -3dBFS (or wherever you decide to put it). I use a relatively fast attack (<5ms) and a slowish release.

If you're tracking (as you should be) with average levels around -18 and peaks a few dB higher, you can see that you should never hit the limiter...but, if you do, you've just rescued your take.

(I'm thinking of vocals on the above...percussion is something different.)

 

[witzendoz]

I have a slightly different take on this.

As a matter of course, I prefer to use no compression when tracking--and dynamics processing is applied after the fact "in the box".

Hay Bobbsy, you missed this line on page one of the thread, Sorry I meant to say that I wouldnt mind limiting the toms in a live situation, not for recording.

Cheers
Alan.

 

[chavernac]

I often use limiters on my drum buss.
LIke actuel limiters with an infinite ratio.

They control the transients. I said "control". you don t actually see ANY gain reduction. But you definitely hear it.
It sounds more controled that way. and punchier.

Try to use your compressor with the fastest attack and highest ratio setting.

 

[Bobbsy]

Hay Bobbsy, you missed this line on page one of the thread, Sorry I meant to say that I wouldnt mind limiting the toms in a live situation, not for recording.

Cheers
Alan.

Ooops, my bad.

Well, at least the instructions for how to set a compressor to be a limiter still apply!

 

[Murdersgalore]

I used to use some limiters during live situations, but then I fired the rest of the band.

 

[Bobbsy]

That's the ultimate limiter! Hmmm...if you gave them the finger as they were leaving, is that a digital limiter?

 

[dainbramage]

If you're tracking (as you should be) with average levels around -18 and peaks a few dB higher, you can see that you should never hit the limiter...but, if you do, you've just rescued your take.

Excuse me, but what?! -18 ?? I usually track as loud as possible, right beneath where my tracks would clip. Am I doing it wrong? Should volume be added later?

 

[aaronmcoleman]

yeah, this has been argued a lot lately. i think just for safety and not clipping it's a good idea to track around -18 to -12 with peaks around -6. There are some physics reasons to stay at lower volume with 24 bits, but i don't know what they are. i mostly just track low to keep from clipping and not have to use a compressor until mixing.

 

[Bobbsy]

There are lots of threads about tracking levels, but, in a nutshell:

The scale used on DAWs is dB(FS) (dB Full Scale) with the 0 level being the point where digital clipping starts. This is very different from the analogue world.

On a mixer or analogue tape recorder, the 0 level on the meters is an arbitrary voltage (generally .773 volts but that's not important). What IS important is that the electronics will have huge amounts of headroom above that 0 level before clipping. Typically, there's something like 18-22dB of headroom.

When you tracked on analogue gear, you'd typically try to have the general level around the 0dB point on the meter with some peaks eating into the headroom so, with the different scales, this equates to around -18dB on the DAW meter (but with peaks going above that).

The advantage of working this way is that, since levels are cumulative, this allows you to mix together multiple tracks without the final mix getting into clipping. If you record every track just below 0dB, you just have to pull down every track while mixing anyway.

The other advantage is that many effects plugins emulate the hardware world and are set up for "analogue" levels--they can get a bit harsh when working near clipping.

The only potential downside of tracking lower is that it puts you a bit closer to the noise floor. However, with any decent gear this should not be an issue. All bets are off with built in gamers cards though! However, since those are also more prone to clipping, they're very much a Catch 22 anyway.

 

[mixsit]

Excuse me, but what?! -18 ?? I usually track as loud as possible, right beneath where my tracks would clip. Am I doing it wrong? Should volume be added later?

The why to that- is to ask why would you run presumably as a methodology, your analog gear (in front of the converters..?) right up to the very edge?
Sure, there's push some stuff hotter' -on some gear = a different sound. (saturation goodies, 'let's go crunchy/thick vs clean/more open').
But as default? -18' (+/-) is 0vu, nominal design level. The other 17 is the analog head room.

 

[Massive Master]

Excuse me, but what?! -18 ?? I usually track as loud as possible, right beneath where my tracks would clip. Am I doing it wrong?

Then you're overdriving the hell out of your front end.

Should volume be added later?

If you're tracking at "more normal" levels, you're probably still going to have to attenuate many sources to avoid clipping the 2-buss.

You just won't have to attenuate them all 15dB or more like you would if you were trying to kiss the red. And it'll likely sound freakishly better because you weren't overdriving your front end.

More if you're bored... And if it isn't in this thread somewhere already... Proper Audio Recording Levels | Rants, Articles | MASSIVE Mastering

 

[rayc]

Massive, I had a few minutes so re read your article. It's still spot on & easily understood (some folk may need pictures!).

 

[dainbramage]

Thanks everyone.

The advantage of working this way is that, since levels are cumulative, this allows you to mix together multiple tracks without the final mix getting into clipping. If you record every track just below 0dB, you just have to pull down every track while mixing anyway.

So are you saying it isn't that important if I record myself playing an acoustic (a stereo pair SDC) and vocals (LDC) at the same time? I mean when there will be no additional tracking.

why would you run presumably as a methodology, your analog gear (in front of the converters..?) right up to the very edge?

Are you saying the gain on my (digitally controlled) preamps should be less but with higher levels in my DAW (which I assume is in the AD) ?

 

[dainbramage]

Then you're overdriving the hell out of your front end.

If you're tracking at "more normal" levels, you're probably still going to have to attenuate many sources to avoid clipping the 2-buss.

You just won't have to attenuate them all 15dB or more like you would if you were trying to kiss the red. And it'll likely sound freakishly better because you weren't overdriving your front end.

More if you're bored... And if it isn't in this thread somewhere already... Proper Audio Recording Levels | Rants, Articles | MASSIVE Mastering

I just read your article, and man was that a revelation!

However I'm still not sure whether I should keep the levels down in my DAW or just in my preamps. I think the preamps would be the important thing to keep lower because as you say in your article analog gear is more like "Perfect, a bit unfocused, a little noisy, "tight" sounding, spectrally distorted, CLIP."

Or should I just keep the volume at lower levels during mixing as well and pray that I will be able to handle proper mastering in the near future.....? I mean, the volume will be really really low at those levels.

 

[Bobbsy]

Thanks everyone.



So are you saying it isn't that important if I record myself playing an acoustic (a stereo pair SDC) and vocals (LDC) at the same time? I mean when there will be no additional tracking.

No. Even there, if you have three tracks running, at at nearly 0dB(FS) then mix them together, the resulting mix will be clipping unless you reduce the levels on all three tracks somewhat.

If you record at the levels being suggested, you can mix all three tracks together and still not clip.

As for where to set the levels, the goal should be to have a consistent gain staging through the system--but this can be easier said than done depending on where you can meter things. Sometimes you just have to take a "best guess" but, where you can monitor levels try to have them consistent in the right range.... -18dBFS with peaks going 6-8dB higher or 0dBVU with some peaks in the red at +6 to +8.

 

[Massive Master]

However I'm still not sure whether I should keep the levels down in my DAW or just in my preamps. I think the preamps would be the important thing to keep lower because as you say in your article analog gear is more like "Perfect, a bit unfocused, a little noisy, "tight" sounding, spectrally distorted, CLIP."

Or should I just keep the volume at lower levels during mixing as well and pray that I will be able to handle proper mastering in the near future.....? I mean, the volume will be really really low at those levels.

First thing - The entire train of thought needs to change. This isn't "low" - This is NORMAL. Tracking "hot" is NOT normal. Mixing hot is NOT normal. As a side-note, the ridiculous levels that most of the stuff "on the shelf" is at these days is also NOT normal. But that's for another thread.

Obsessing about volume is the fastest route to not having any when you want it anyway.

Hopefully this link still works -- (yep) -- Open this link in your favorite MP3 player. http://www.massivemastering.com/special/LiveRecording.m3u

It's a "snippet" file -- Little chunks of a bunch of tunes recorded live. And I mean freaking live... "Hey, can you record this for us?" at the last minute live. "Aux 3 feeding L and aux 4 feeding R post-fade from already set input levels" freaking live. No panning (except the relation from one aux to the other), nothing more special than "GC" quality gear in the chain (57's, 58's, MD504's, SM81's, BBE direct boxes) running into an old, noisy A&H 2200. "Best guess" on everything as I had to set levels live, in the room, with open-backed headphones. The bass was too loud in the house (so it's a bit quiet on the recording) the hat was blaring through the wedge so much that you're hardly hearing the actual hat. No EQ, no compression, a bit of reverb sent down the same pipes (aux 3 and 4). And yes, some conservative tweaking here when done.

Okay - What's my point? I literally had "real time" on one short soundcheck tune to dial everything in. I was running -10 outputs into +4 inputs on a converter set I never used for this sort of task with no chance to calibrate and no chance adjust once everything was up and running.

THE SHOCK: The top peak on the entire recording was -42dBFS. That's the top peak of the sum of the elements in the mix. Figure each individual track would have been running around -65 to -60dBFS.

THE ASSESSMENT: Is in noisy? No (and believe it or not, I ran this in a layback pass to tape during processing). Is it clear and clean? Yes. Is there focus? Yes. Can you stand back and literally point to each element in the mix? Yes. Does it have a reasonably acceptable playback volume without sounding squashed? Yes.

THE GUILT: Now - Did *I* screw up? You bet. (A) I was too conservative with levels (which doesn't really bug me - Wish I had a few more minutes that's all). (B) Didn't realize that the outputs of the auxiliary sends were at -10 (or the top peak would've been closer to -30dBFS, which would've made me much happier).

But with everything stacked up against getting a listenable recording as far as levels are concerned, do I think that getting hotter levels would have made an appreciably better sounding recording? Hardly. And even after adding 40dB of gain just to bring it into reality, I'm still really happy that the levels were too low (this WASN'T "normal" - This was truly "too low") than too hot.

Keep a decent amount of headroom at every conceivable point and every possible phase in the process. That's normal. Work within the specs of your gear. That's normal. You're going to wind up with mixes that are considerably lower in playback volume than the typical "on the shelf" commercial recordings. THAT's normal. And those recordings will handle the "abuse" of eventually being brought up to those ridiculous levels far better and with far less damage than they would otherwise.

I wish that last part wasn't normal - But again, that's for another thread.

 

[snow lizard]

John, that's a great story. The clip sounds fine in terms of noise floor - one of the only sane arguments for pushing tracking levels up - hereby rendered moot!

I love this!

Seems like all the sounds have a focus and ability to intertwine that gets destroyed when the tracking levels are too hot.

Can you comment on the converters and word length used?

 

[Massive Master]

Apogee MiniMe @ 44.1/24 -- Wish I still had it - That was a phenomenal little box.

Probably should've specified the 24-bit part... This is one of those things that would have sounded significantly different had I been working in 16-bit. There would've been dither noise sitting somewhere around -70dBFS (which would have been -30dBFS by the time I added 40dB of gain), the noise floor itself (-96dBFS) wouldn't have been low enough to capture the low-level parts.

But in 24-bit, even those peaks around -42dBFS had higher theoretical resolution and better dynamic range than a 16-bit recording at full-scale.

I'd imagine at some point some astute reader is going to ask about the front-end gain staging also -- And there's a point to be made there as well.

The setup was for a concert. A ridiculously overpowered system (nearly 15kWRMS in a 450-seat house) kept the input gains very low... Much lower than normal. The metering on the thing was nearly non-existent, but I'd figure the typical preamp was running 20dB or more below line level --- Each individual input running a full 20dB or more lower than where it should've been. Similarly overpowered monitor amps (somewhere in the 2kWRMS range) still allowed for solid wedge levels even with the way-below-line-level input.

What the place needed (and what I eventually had Crane Song craft for me) was a set of input pads for those powered Meyer mains...

But for clarity -- This wasn't an occasion where I was getting normal input levels and then attenuating, then bringing back up later. This was something with abnormally low voltages from stem to stern.