Mastering Limiters

Nola

Well-known member
What do mastering limiters do to sound other than making it louder?

e.g. say you have a distorted guitar and run it through a brickwall limiter. IME it makes it more brittle and crunchy compared to the raw mix. I'm just wondering if it's normal the limiter changes the sound, and if so, how it changes sound and what the best practices are. Is there a thread that lists best practices or a book someone can recommend?
 
That's what I hear as well--brittle, "crunchy" if you like but not in a good way. Then again, we are probably only noticing this effect where the limiting was done badly.
 
I literally just posted this which is probably more info than you wanted and likely doesn't answer your actual question. :)


Edit - fixed link
 
Last edited:
I literally just posted this which is probably more info than you wanted and likely doesn't answer your actual question. :)


Edit - fixed link

it was interesting to read, but i'm not sure it explains what a limiter actually does to the wave form. is it just cutting off peaks and raising the quiet parts? Does this combination create brittleness somehow? Robus said it's only if limiting is done badly. I just used a preset to see what would happen, and it sounded brittle and crunchy, but how does someone learn how to limit well? Is there a book or article on mastering limiters and how to use them correctly?
 
Maybe it's best to post what I do and see what's wrong, if anything.

1. usually i record -18dbfs
2. i'll use minor compression, if any. e.g. distorted guitars i rarely compress. an electric clean guitar i might compress a little so it's like -3db or so. then use makeup gain to get any volume i need.
3. vocal i'll compress a little like -3db to -5db, than make up gain.
4. eq everything to sit where it sounds good.
5. mix it down.
6. run it through some pluging brickwall limiter.

What would I do in this process to make it where I don't have to use much limiting to get adequate volume? i like average volume around -19 to -16db, which i guess is low by today's standards but sounds normal to my ear.
 
it was interesting to read, but i'm not sure it explains what a limiter actually does to the wave form. is it just cutting off peaks and raising the quiet parts?
No. It's cutting off the peaks. If you/it also adds gain- then you are bringing everything up [not just "the quiet parts"- everything.]
The more you raise something into a hard limiter (or the limiter does this, or you lower it's threshold down into it to do it) ..the more something like this distorts the program.
Does this combination create brittleness somehow? Robus said it's only if limiting is done badly. I just used a preset to see what would happen, and it sounded brittle and crunchy, but how does someone learn how to limit well?
Yes. First of all "something like this".. is anything w/ a very fast attack and release alters waveforms too fast not distorting them.
2nd- what you're limiting. A distorted guitar is already 'clipped and distorted. All compressing to a large degree and certainly limiting can do [here] is... make it more distorted. Along with bringing up other nasties that used to be down lower in the track. (i.e. if you liked this clipped guitar tone- why would you distort it some more.
Save the limiter for the short clean spikey things that can stand a bit of clipping w/o Sounding Like Ass [a technical term :guitar: Do it until they do S.L.A, then back off. That's how far you get to go B4 S.L.A.
 
Really aggressive limiting turns into distortion real fast. Distortion adds harmonics - multiples of the original frequency. Since a lot of the peaks that (should) get clipped at this point are pretty fast transients, they usually already have quite a bit of high frequencies, and then you multiply those frequencies and get even higher frequencies, and those frequencies often end up being higher (in frequency, not necessarily volume) than anything else in the mix. In a very real way, the whole spectrum of the mix gets tilted a bit toward the treble end just because we're adding more crap up there.

Then too if it's like a kick drum that's pushing it over the threshold, the low frequencies in that sound aren't allowed to be as much louder than everything else, so it's almost like losing some bass at the same time.

And that's not to mention aliasing which happens when multiplying the original frequency puts it over half the sample frequency and it "reflects" back and ends up not being actually harmonically connected to the original sound anymore. Decent mastering limiters work hard to avoid that, but there's only so much you can do, and some of what you have to do sort of undoes what the limiter is trying to do.

Anyway, there's like three active threads on how to reduce dynamic range without destroying the mix. Maybe instead of making us type all that stuff again, you could read them?
 
One of my buddies sometimes has an unbelievably spiky attack on his vocal. Until recently reading of some using a limiter I never would have considered hard limiting.
PSP's Xeon worked nicely w/o inflicting any weirdness.
Thing does have sort of a two-stage attack with a release you can slow down, might be part of it.
 
Yes, limiting can make a mix sound brittle. It does it to the whole mix, not just the guitars. But that is where it's most noticeable.

If you are trying to get an average volume of -16dbfs and the mastering limiter is changing the tone of the guitars, the mix has problems. (in fact, I can't imagine a scenario where that's possible) I can usually get a guitar heavy band mix to -12dbfs without any master limiting on the mix. (I just normalize the mix to peak just below 0dbfs)

The trick is to control the dynamics of the individual instruments within the mix. Then, when you balance everything, there isn't really much for the limiter to do, besides beat back some of the drum transients.
 
thanks farview. for some reason the stereo mix coming out is like -28db. i'm not sure why it's so low. i think b/c i keep my programmed drums at unity always [on the channel fader] and base everything around that. so my bass might be like -12. everything just seems to get pushed down once i set the programmed drums. :/

i'd love to get to where i don't need to limit b/c my mixes sound so much better before i put them through the limiter. i am doing something wrong but not sure what.

do you watch the db on the master fader when mixing and push the main instrument to a certain [i.e. loud] level then fit everything around that? i'm just wondering how to get it louder in the mix stage and what the heck i'm doing wrong.
 
do you watch the db on the master fader when mixing and push the main instrument to a certain [i.e. loud] level then fit everything around that? i'm just wondering how to get it louder in the mix stage and what the heck i'm doing wrong.

Sort of, yes. These days I generally build a mix around kick, snare and lead vocal. I'll start with kick and snare, setting them to peak around -12dBFS, get the vocal to sit with that, add the bass and then hang everything else on that backbone. But I don't do it with the channel fader, I use channel gain which precedes all the channel inserts and fader.
 
Get the drums so that they peak around -6 or so, then mix around that.

Once you have it mixed, normalize it and see where you are at volume wise.
 
thanks boulder and farview. i'm going to try those tonight (i have jury duty in a bit ugh)
 
I use channel gain which precedes all the channel inserts and fader.

i never really thought to do that. for some reason i thought raising either the channel gain or the channel fader would increase noise floor and start creating distortion, which is why i always just started with drums at unity and built around that.

i think this idea could be a breakthrough for my mixes b/c i really want to use less compression/limiting and if i can turn the gain up (you mean the gain/trim on the mix board right?) then that's great.

do you only use the faders for small fine tuning type movements after you get the gain knob where you want it?
 
i never really thought to do that. for some reason i thought raising either the channel gain or the channel fader would increase noise floor and start creating distortion, which is why i always just started with drums at unity and built around that.

i think this idea could be a breakthrough for my mixes b/c i really want to use less compression/limiting and if i can turn the gain up (you mean the gain/trim on the mix board right?) then that's great.

Whatever the noise floor of your signal is once it's digitized it will rise and fall with any gain changes applied whether that's right up front (DAW track gain) or later in mastering. The gain I meant above is in the DAW, but if you can get the right levels in analog so it's right in digital then adjusting gain in software isn't necessary.

do you only use the faders for small fine tuning type movements after you get the gain knob where you want it?

I use the track gain to get things in the ballpark and then use the faders for conveniently making smaller changes. So I guess the answer is yes.

For vocals I often make many DAW gain adjustments to even out a track's levels, then compress, then automate the volume. I rarely automate drums and bass.
 
Modern DAWs have an ungodly amount of dynamic range available. There is a real noise floor and there is a ceiling, but they are so far out of the range of even pretty severely abnormal use that we really don't have to worry about them. -28dbfs is pretty low, but it's still hundreds of db louder than the internal noise floor. 0dbfs is as loud as your converters will go, and the loudest thing that will get rendered when you go to a fixed-point format for distribution, but your mix bus can get literally billions of times louder than that without actually clipping internally. There's a lot of room for error.

With that in mind, I think you might see that the actual level at the master fader doesn't really much matter. If it's peaking way below 0, turn up the fader. If it's over 0, turn it down.

What you should be looking at more when getting to the final stages of a mix is the dynamic range of the signal - the difference between the loudest peak and RMS average. That is ultimately what is going to come across as loudness, and the point of compressor and/or limiter on the mix bus is to make that number smaller. Once you get your DR to within the range you're shooting for, you then basically can adjust the fader so that those loudest peaks hit just a little below 0. I shoot for -0.6 or so, but I think a lot of folks go for -0.3. It's not super critical, but it's nice to leave room for "inter-sample peaks" and/or mp3 fuckification.

That's not to say that you shouldn't shoot for "proper gain staging" starting at the track level. It can make things just plain easier, and some plugins - analog simulations etc - can be calibrated to work best at some certain level. But it's not always a deal breaker, and the faders in your DAW aren't going to add either noise or distortion on their own.*


* There are a couple of DAW packages out there whose internal mix structure is essentially analog emulation. I've never tried one of those. One would hope they wouldn't add noise at the bottom end, but they will saturate/distort if you push the levels into them in much the same way an analog mixer might. Those are the exception rather than the rule, and I'm pretty sure that if the software you use offered that "feature", theyd use it as a selling point, and you would probably know.
 
Since you are using programmed drums, it sounds like the velocities of your midi track are pretty low. If they are sitting in the 60 range, your hits are about half volume. Also, if you are using a sample set that triggers different samples at different velocities, you are triggering weak hits.

If you are using loops, just turn them up to get them to a decent level and build the mix around that.

Without hearing what you are doing, it's hard to get an idea of why your mixes are so quiet.
 
I always find the sampled drum kits to be too loud, out of the box...I mean, some of the Toontrack packs are just slamming if I say....just pull a preset groove and drop it in the MIDI Editor.
I'm not complaining though...which means the drums were recorded nice and hot...and turning shit down is always a lot easier than trying to get all the levels up.

But yeah...like Jay said...if your velocities are only averaging in the 60-70 range...that's not going to hit the meters too hard at all. I think for loud sections, like choruses in more uptempo, livelier songs...my Kick/Snare velocities are anywhere from the low one hundred teens into the one hundred twenties...but I generally avoid using 127, the max, except on occasional hits.
If you listen to the sound...there is often a distinct difference between 126 and 127 on a lot of sample hits, and the 127 tends to get too crisp, where 126 will have the solid velocity, but the sample tone stays nice and fat/chunky...at least that's what I've noticed.
Anyway...didn't mean to segue off on a some drum sample velocity thing. :)

So yeah, like some have said...once you get the level of the drums and the vocals set...everything else should follow that.
I think some people will focus too much on guitar track levels first...and that might make it more cumbersome to get a nice balance and solid overall mix level. You end up pulling things up/down too much if you don't let the drums set the stage so to speak.
 
Back
Top