Compressor attack setting

davecg321

New member
I am using the waves ssl to 'glue' my track. I'm primarily mastering (trying to at least) my own acoustic music.

Up until now I had been using an attack setting of 1ms.

After lots more reading on the subject (acoustic mastering) 1ms may be somewhat heavy handed. Using a slower attack definitely 'opens' up the sound more and is more natural.

However, using a slower attack catches less of the transients which cause problems when I'm trying to get my limiter/rms meter to sit where it needs to be.


Yes, I then go back to my individual tracks acoustic gtr/vox/piano/upright bass and try to tame them more with compression/mixing but again I'm forced in this way to be too heavy handed.


Where should I set the attack on the master bus compressor and does anyone have have any further tips with regardsto getting things under control

Many thanks

D
 
Using a slower attack definitely 'opens' up the sound more and is more natural.

However, using a slower attack catches less of the transients which cause problems when I'm trying to get my limiter/rms meter to sit where it needs to be.

There are surely more reliable people here for mastering q's, some of which I even rely on myself for master bus q's, but I'll chime in on this one.

Your dilemma, as quoted above, pretty much sums it up. If the fast attack kills too many transients, and the slow attack lets too many through, then you may want to try a compressor that has a dial where you can put in the exact milliseconds. That way, you can find a balance you're liking. Don't forget that release may be just as important as the attack, if not more.

A 1ms attack isn't going to "glue" something as much as it will dull it by killing the transients, but that all depends on what your release is set to. A fast release will let go of the compressor and it may grab that transient anyway. A slow one will def kill the transient. For a typical "glue" compressor setting, you may want to try a med-slow attack with a long release. That aims to keep the track in a range of db's, sort of, and almost holds it there throughout.

This is all very dependent on the mix and the other elements at play, though.
 
The ssl bus compressor does a really specific thing. If what it doesn't isn't working for you, you may want to try a different compressor.

That said, if you get the "glue" with the slower attack time, then you will have to rely on the limiter to tame the transients and get the rms level up.

Also switch the ratio, that compressor reacts differently at different ratios. The knee get changed.

It might also be worth slapping compressors on your sub groups. A lot of little bits of compression can sound better and more natural than one compressor trying to beat everything down.
 
Yes the ssl compressor works great and is simple enough for me to use.

I love the sound of a slow attack of 20 ms (haven't played around too much with release yet)

Ratio is set to 2:1

I then finally use a limiter last in chain to boost the signal but can only go so far until I get intersample peaks and way too much gain reduction. On that note, is it OK to apply constant gain reduction on the limiter of say 3-4 db? Or is GR at this stage only intended for peak transients?

D
 
As long as it sounds good, you can do what ever you like. But there is a limit to how much you can compress and limit something before it falls apart. Compressing or limiting individual tracks or subgroups in the mix will keep the final limiter from having to catch big transients, which is what will make things sound bad.

As for the slower release times, you will kind of lose the glue effect, and just make the compressor ride the volume.

Increase the ratio and see what it does for you. The attack and release react differently at different ratios.

But either way, a little compression of individual things or groups will help you get the final volume up, without mounding squashed.
 
Atack time doesn't mean a whole lot on its own as all the parameters along with your signal are going to be interactive. You can slow the attack time until you notice the natural attack and much of the high frequency content return to the signal. Blend to taste. If you're not getting enough "squish", it isn't really a "glue" issue. What the exact number ends up being is irrelevant compared to what your ears tell you, and it will change on a case by case basis. Maybe lowering the ratio and threshold and aiming for modest gain reduction will give you more of a natural " glue" effect without killing the transients.

Different compressors can vary quite a bit in what they can do to your signal. Some are clean, some are crunchy. Both types have uses. Some types will only give you "drive" and "output" controls with perhaps a few other bells and whistles, where it's all compression all the time, but the ratio increases with the signal strength. (you can simulate this effect by runnning 3 or so instances of a compressor in series at light gain reduction settings, and progressively higher thresholds - makes a "soft knee" effect) Some of these can do fairly heavy gain reduction with minimum artifacts, as opposed to a brick wall limiter.

If you have a few different ones available to audition, I'd try that and push them to extremes to find out how clean or crunchy they are, and where things fall apart. Having a dizzying array of compressors to use isn't extremely helpful, but settling on a couple of different ones that you like, that do very different things, can be.
 
On the SSL bus comp that he is using, the attack, release and ratio controls are stepped, there are no in-between settings. Also, that compressor changes the knee when you change the ratio. Higher ratio, softer knee. So this is definitely a compressor that you have to ignore the numbers and just listen.
 
The ssl bus compressor does a really specific thing.

What specific thing does it do? I always see the SSL channel strip plugins but never feel the urge to get one.

OP, if transients are the problem address them in the mix. You can go in an manually reduce the really bad ones. To me, mastering shouldn't involve much at all. Less is more. Maybe a few db of gain reduction with a compressor to glue it, then get your volume where you want, then you're done. All of that varies by mix so you have to develop the ear. I wouldn't do more than that at home as an amateur b/c beyond that you probably do more damage to the mix than good. I kinda sucks that home recorders have to be musicians, mixing engineers, and now mastering engineers. Too many hats and it keeps us from making more tunes. I kinda hate that part. Sorry, tangent.
 
Ya, there is no need to struggle with a glue compressor when you can see transients are giving it grief. Isn't that kinda' normal track level stuff before one even starts mixing ?
 
Oh, and imo the most important part of mastering is getting a cohesive sound across an album. Volume should be looked at in that regard. Like no big volume differences between the songs, but volume shouldn't be confused with mastering. Somehow mastering and volume seem to mean the same thing these days. If you have an album of 10 songs and the RMS is like -18db across all of them then that's fine even if it's quiet (that was pretty much the standard throughout the 60s and those are all great records). People can always turn up a system to listen to an album, and it will probably sound better anyway than something squashed to -10db with a limiter's preset or some vague idea of how the limiter works.

Sorry, if I'm stepping on any mastering engineer's toes those are just my opinions and advice from my own experience.
 
garww said:
Ya, there is no need to struggle with a glue compressor when you can see transients are giving it grief. Isn't that kinda' normal track level stuff before one even starts mixing ?

Either that or it's normal steady state/apocalypse RMS levels on acoustic sources stuff. I'm not sure what kind of RMS levels we're talking about here.
 
Either that or it's normal steady state/apocalypse RMS levels on acoustic sources stuff. I'm not sure what kind of RMS levels we're talking about here.

Ya, we don't know the character of it. Over here, I can often get away with just deleting the bad spikes
 
What specific thing does it do? I always see the SSL channel strip plugins but never feel the urge to get one.
The SSL bus compressor is different than the SSL channel strip compressor. The bus compressor is really made to glue a mix together, or make a drum bus pump. it doesn't react like a normal utility compressor would. It's its own thing.

I use the channel strip all the time. I like the EQ, you can add a ton of high shelf without it getting spikey and harsh. But also, it's the sound of the stuff that I really liked from the 80's and 90's. I actually use the gates because they react the way I expect them to. It's just the tool for the sound I tend to go for.
 
I've referred back to my mix to try and tame the transients more (guitar and vocals) but I'm already getting 4db of gr on guitar and 6db or gr on vocals....

?
 
That's OK, as long as it still sounds good. And that is 4-6db that the bus compressor and limiter doesn't have to catch.
 
I've referred back to my mix to try and tame the transients more (guitar and vocals) but I'm already getting 4db of gr on guitar and 6db or gr on vocals....

?

That's nothin'. Depending on the vocal I can get into double digits of reduction. Usually that means it's time for manual editing, but sometimes I just go for it.
 
I had this exact problem on my drum tracks, and posted a very similar question (in fact, I had to double take as I thought I was reading back my old post for a moment when I read your question).

The problem as I see it is that the transients for the drums were spiking, but what I discovered is that the spike itself was not meaningful sound. So I put a limiter (L3) on my snare and bass tracks. I tried multiple compressors, look ahead compression, etc. - but the problem is that they all would attenuate the overall signal, and I would lose the punch of the transient. When I went to the limiter, it was basically chopping off the spikes (very impressed with the L3, when I do the same with the SoftClipper, it distorted the sound), and it turned out those instantaneous bits of the transient were not important, or even audible actually. The meat of the snare and bass hits is what I wanted, and I was able to turn those up louder to get the beef I wanted without jumping through compression hoops.

My method was met with some appropriate scrutiny, but I got the sound I wanted (well, except for one recently posted track, but that was nothing to do with this technique). I was able to preserve the instantaneous POP of the transient, but actually keep the overall range in check, which allowed me to still get strong volume at the mastering stage when I squeeze the whole mix (and now, I only had to do so very very little).
 
Back
Top