What does an attack parameter do in a brickwall limiter?

[manducator]

Hello,

Some brickwall limiters (like Fabfilter Pro-L, which I absolutely love) have an attack time parameter. But what does it do on a brickwall limiter?

I know what the attack time on a compressor does (it lets the transient pass, before the compression kicks in). But how does the attack time parameter work on a brickwall limiter as it isn't allowed to let a transient bypass the threshold anyway?

[mystran]

No idea what it does in Pro-L, but you can certainly modify attack if you have some lookahead. As long as your lookahead is at least as long as the attack you can still hit the necessary reduction in time.

[bmanic]

You might want to the manual of Pro-L, it explains the attack and release knobs.

Basically Pro-L (and other advanced limiters) have several program dependent "envelope" followers in the algorithm. Attack in Pro-L controls the "blend" of the fast and slow release envelopes. If you set attack all the way to zero, only the slow envelope follower will be in use. If you set attack to maximum, then Pro-L acts almost like a clipper, almost completely bypassing the slow envelope part.

It's a bit more complex than that inside the algorithms but that's close enough to an explanation how and why you would want to use the attack and release controls.

They can also be used to "shape" the response of the limiter and thus how it affects the music. Sometimes it is NOT desirable to get a too punchy sound but rather you want to squash the transients a bit and keep the limiter slow. In this case you set the attack know to some fast values (like 0.5 to 5ms) and keep release fairly long (like 900 to 2000ms). The opposite would be a highly aggressive limiter sound, setting attack to large values and release to low values.

Cheers!
bManic

[SJ_Digriz]

I know what the attack time on a compressor does (it lets the transient pass, before the compression kicks in)

Your definition is not really correct, which is probably why you don't think it makes sense on a "brick wall" limiter value.

Although it is implemented a bit differently from compressor to compressor, in general the attack is the amount of time after the threshold (if it has one) that it takes to reach the ration. Even on a brick wall limiter.

If you think about all of the possible ways to deal with that threshold, then you'll see why, strictly speaking, it isn't really about letting a transient pass. Although in practice, that's what usually happens.

[contrary]

Instantaneous attack would result in a thing called waveshaping.

A little something to think about ; what's the difference between hard limiting and soft limiting ?? warming and clipping ???


Find some time and read repeatedly ...


http://www.simulanalog.org/clip.pdf




Have fun

[manducator]

You might want to the manual of Pro-L, it explains the attack and release knobs.

Oops, that reminds me to the fact that I started to use Pro-L after seeing fabfilter's tutorial and indeed, forgot to read the manual, which will be more in depth.

I think Pro-L is the only plugin from which I haven't read the manual.

Your definition is not really correct

Ok, so does it mean that playing with the attack time, changes the attack curve, to reach the desired ratio? Is the attack time some kind of 'curve stretcher'?

Btw, Fabfilter's Pro-C's manual does state this:

The Attack knob sets the time after which gain reduction sets in.

Isn't that the same as I wrote? Mmh, maybe not, reaching GR is something different than I wrote, I guess.

It seems that I don't uderstand what a limiter does. I thought I knew... If anyone has more info, I'm happy to digest.

http://www.simulanalog.org/clip.pdf

Thanks. I'll dive into it but it looks 'heavy'.

EDIT: I do understand right now, after taking a look at the manual that the attack changes the time on which the release starts. So the attack time changes something about the release, that was confusing but I start to get it.

[bmanic]

The attack knob varies the time for when the 2nd release stage (!!) sets in. It's the slow release stage of the limiter, the one that "follows" the music.. ehm.. musically..

Cheers!
bManic

[manducator]

Ah, it's starting to make sense to me, bmanic.

[xx JPRacer xx]

I'm not too familiar with this either. But reading your last post bmanic the way I understand it is the attack time be considered a sort of "hold-time" before the release phase start? (presuming the volume as returned below treshold)

[bmanic]

No, it's not a hold time nor can it be considered a "sort of" hold time. Attack and release cycles happen all the time, constantly, in some cases even clipping (meaning instant attack and release). The attack knob just sets the approximate (it's all very program dependent) time for the SLOW release stage to kick in.

Most modern brick-wall limiters consist of several stages within the algorithm.. some even contain a whole bunch of stages, all interacting with each other.

I can not go into exact details obviously as that would break the NDA I've signed with FabFilter.

Cheers!
bManic

[xx JPRacer xx]

Ok, thanks for your explanation, I'll try to read more on the subject. But reading your post now I understand that these settings are aproximations, in my head these were absoulute values and I didn't know about these "multi-stages" algo. Thanks.

[bmanic]

Jep. I'm actually a bit bummed that FabFilter didn't set attack and release knobs labels as a percentage in stead of the current ms value. Attack knob at 10 000ms (!!) looks ridiculous but is very useful. It turns pro-l into a program dependent clipper. Attack 100% looks much better imho and is way less confusing.

Cheers!
bManic

[SJ_Digriz]

Ok, so does it mean that playing with the attack time, changes the attack curve, to reach the desired ratio? Is the attack time some kind of 'curve stretcher'?

You are looking for a single yes/no answer where there isn't one. As bmanic and I have pointed out, there are a zillion implementations of compressor/filter concepts. The only constant concept for attack is the amount of time to reach the next phase of the effect once an initial threshold has been reached. It can be curve based, it can be controlled by LFO, it can be flat, it can respond to circuit design principles etc. The device/program will determine how the attack works between start and the next phase of the effect. But it is almost always a time based function. Although, as you have probably guessed, you can have an attack that responds differently to incoming amplitude/velocity/control etc..

[softegg]

I just realized that I really misunderstood what the attack knob in Pro-L was doing.
This thread turns out to be very useful.