The dance/pop mix/master comp/lim THEORY thread

[Apollo Soul]

Thank you, all for indulging me in my latest obsession, but whenever I or anyone asks about a buss comp, it generally gets a round of everyone just suggesting whatever they use... and invariably, every box ever made gets suggested for every use ever imagined.

So... let's take a step back. Forget the specific models, etc... I'd like to talk about the fundamental concepts and see if it helps me to arrive at a mix/master CHAIN that is truly optimized for dance/pop.

Let me be clear... I'm talking about techniques for mixing and mastering dance/pop here with the general underlying assumption that it will need to be relatively loud. Not interested in debating the premise... just laying the foundation. I am also assuming that the music will be compressed as a whole... that is, master buss compression and/or limiting.

So, here's what I've learned so far:

If you have a variety of elements stacked, each may have it's own transients, and those transients will cause compressors and limiters to kick in.... if they aren't dealt with on the individual tracks, then the master buss compressor will eventually have to deal with them.

Most master buss compressors only sound particularly good within a relatively small range. If there are massive spikes being fed, it is harder to dial in that range, and more of the music will be outside the range.

THEREFORE:

I believe it's much better to deal with the dynamics PER TRACK before mixing. Otherwise, the kick's attack, snare's attack, guitar's attack, etc... could all be stacking and feeding the MB comp a peak that's 20db hotter than the next beat... and what will happen is it will clamp down hard. If it doesn't clamp down hard enough, remember, this is getting maximized at some point, so another limiting stage or even digital brickwall will need to be employed to deal with the peak.

Unfortunately, that extreme measure will screw up the whole mix as all the elements jump unnaturally.

I have found it's much less damaging to deal with dynamics per track so that there are no such massive spikes on mixdown, and the MB compressor can reside comfortably in it's sweet spot.

I know... a lot of typing, but bare with me, I'm going somewhere with it.

So... assuming you are working this way, there are no big spikes... yet there are still peaks because there are moments where a vocal, kick, snare, etc... happen simultaneously vs other times when they don't.

Now, using the typical tools associated with dance music like ssl, c2, obsidian, The Glue, etc... the general formula is to have a longer attack time so the transients come through... and then the release may or may not pump with the music.

So... taking the simplest example of house music with a 4 on the floor beat, each beat creates a peak... and the peaks aren't too far from eachother. What this means is that a lot of the advice out there about limiting doesn't work because if you apply a limiter, you won't get too far into the max GR before it starts affecting EVERY beat... perhaps only 3db or so at the max peak before it starts reshaping each and every hit.

Limiting beyond that is likely to just wreck the drum sounds and sound bad.


So... here's what I'm thinking. Might it make more sense to use the typical ssl type tools to get the type of pump you want...

... and then rather than pushing the limiting to the point where it limits every beat, use it very sparingly...

... and now here's the moment of truth.

It's still not hot enough. If you don't do somehthing else, it's going to end up getting something severe like a digital brickwall. What if you want to avoid that? What if the GOAL is to get the density up with as little damage as possible. Pump if you want it from the ssl for style, yes... but after that, to put an absolutely invisible squeeze on it.

I'm thinking in theory at least, it would be better to have a low ratio compression do the entire waveform. The balance of everything remains exactly the same... just half the amplitude... or whatever ratio you have... but none of the relationships change. There's no weird pumping because there's no release. There's no change in the drum sound because there's no attack.

Am I off on this? I feel like this is the general path I want to investigate, but I don't know of any such beast... except maybe the HOLD on a couple ancient pieces.

Is there a way to just completely evenly divide each and every sample by a certain ratio?

Wouldn't that keep everything exactly intact except for narrowing the dynamic range?

If so, is there a way to actually do this? What's the right tool for the job?

LOVE to hear some discussion on this... or other specific ideas as to how and why it makes sense to use a particular chain in a particular order to get the maximum loudness with minimum damage when dealing with big bad electronic kicks.

thx.

[hive]

Originally Posted by Apollo Soul

Thank you, all for indulging me in my latest obsession, but whenever I or anyone asks about a buss comp, it generally gets a round of everyone just suggesting whatever they use... and invariably, every box ever made gets suggested for every use ever imagined.

So... let's take a step back. Forget the specific models, etc... I'd like to talk about the fundamental concepts and see if it helps me to arrive at a mix/master CHAIN that is truly optimized for dance/pop.

Let me be clear... I'm talking about techniques for mixing and mastering dance/pop here with the general underlying assumption that it will need to be relatively loud. Not interested in debating the premise... just laying the foundation. I am also assuming that the music will be compressed as a whole... that is, master buss compression and/or limiting.

So, here's what I've learned so far:

If you have a variety of elements stacked, each may have it's own transients, and those transients will cause compressors and limiters to kick in.... if they aren't dealt with on the individual tracks, then the master buss compressor will eventually have to deal with them.

Most master buss compressors only sound particularly good within a relatively small range. If there are massive spikes being fed, it is harder to dial in that range, and more of the music will be outside the range.

THEREFORE:

I believe it's much better to deal with the dynamics PER TRACK before mixing. Otherwise, the kick's attack, snare's attack, guitar's attack, etc... could all be stacking and feeding the MB comp a peak that's 20db hotter than the next beat... and what will happen is it will clamp down hard. If it doesn't clamp down hard enough, remember, this is getting maximized at some point, so another limiting stage or even digital brickwall will need to be employed to deal with the peak.

Unfortunately, that extreme measure will screw up the whole mix as all the elements jump unnaturally.

I have found it's much less damaging to deal with dynamics per track so that there are no such massive spikes on mixdown, and the MB compressor can reside comfortably in it's sweet spot.

I know... a lot of typing, but bare with me, I'm going somewhere with it.

So... assuming you are working this way, there are no big spikes... yet there are still peaks because there are moments where a vocal, kick, snare, etc... happen simultaneously vs other times when they don't.

Now, using the typical tools associated with dance music like ssl, c2, obsidian, The Glue, etc... the general formula is to have a longer attack time so the transients come through... and then the release may or may not pump with the music.

So... taking the simplest example of house music with a 4 on the floor beat, each beat creates a peak... and the peaks aren't too far from eachother. What this means is that a lot of the advice out there about limiting doesn't work because if you apply a limiter, you won't get too far into the max GR before it starts affecting EVERY beat... perhaps only 3db or so at the max peak before it starts reshaping each and every hit.

Limiting beyond that is likely to just wreck the drum sounds and sound bad.


So... here's what I'm thinking. Might it make more sense to use the typical ssl type tools to get the type of pump you want...

... and then rather than pushing the limiting to the point where it limits every beat, use it very sparingly...

... and now here's the moment of truth.

It's still not hot enough. If you don't do somehthing else, it's going to end up getting something severe like a digital brickwall. What if you want to avoid that? What if the GOAL is to get the density up with as little damage as possible. Pump if you want it from the ssl for style, yes... but after that, to put an absolutely invisible squeeze on it.

I'm thinking in theory at least, it would be better to have a low ratio compression do the entire waveform. The balance of everything remains exactly the same... just half the amplitude... or whatever ratio you have... but none of the relationships change. There's no weird pumping because there's no release. There's no change in the drum sound because there's no attack.

Am I off on this? I feel like this is the general path I want to investigate, but I don't know of any such beast... except maybe the HOLD on a couple ancient pieces.

Is there a way to just completely evenly divide each and every sample by a certain ratio?

Wouldn't that keep everything exactly intact except for narrowing the dynamic range?

If so, is there a way to actually do this? What's the right tool for the job?

LOVE to hear some discussion on this... or other specific ideas as to how and why it makes sense to use a particular chain in a particular order to get the maximum loudness with minimum damage when dealing with big bad electronic kicks.

thx.

good sounding analog clipping is the most real, punchy sounding way of reducing dynamic range in my experience. best reason I've found to stem out to an analog desk is to take the dynamic range on my drum buss down a bit by hitting red on the channels. works great for more aggressive types of music. can be done with plugin saturation as well of course but the real thing does sound more, erm....real.

add to that equation, mixing your stems a bit on the hotter side to saturate the console's mixbuss and end up with even more natural sounding dynamic range reduction. at the end, if done properly, you get a hotter master with plenty of vibe and not needing much brickwalling at all.

[yotonic]

Hive nailed it. And it sounds especially good on House music, just listen to any of the good stuff from the eighties and early nineties, Paul Johnson, The Goodmen Fresh Fruit, whoever. All of those guy were slamming drums in a desk, a lot of old school House producers still keep those Mackies around to do just that, I know.. It's that analog clipping on drums that makes old house sound so thumpy and sexy.

Especially with a genre that is all about the drums and instrumentals. I personally can't stand what digital has done to House music. As a club owner for over 15 years I have heard the whole evolution standing on the dance floor, from analog and vinyl to digital, laptops and CDs. While some of the writing has improved the sound has not. And for House Music the "feeling of the sound itself" is so important. Especially with such a repetitive, instrumental genre blasting at you from 15 feet away.

The late eighties and early nineties were an amazing time for House and House DJs right along with the birth of Boutique Hotels, and an a#$load of money flowing out of Wall St. And it was before 9/11 so there was a lot of rolling going on which is "analog clipping at the source".

[Apollo Soul]

hmmm... ok. Thx.

Seems in line with what I've been finding... that density needs to be achieved before the mix phase.

Setting aside a few things for the moment personally... like the fact that I'm all ITB except for some comps, and that mixing to stems would be a major workflow killer without some kind of multi-out or summing mixer...

Someone had suggested printing the mix to analog tape. Unfortunately, that's the ultimate workflow killer since you don't get to hear your mix back until after the tape... unless you've got a setup where you can monitor off a playhead after the record head. It's been a ling time since I've used tape, and when I did, it was used for tracking, not this way, so I don't remember if that was doable.

So... first of all, I'd like to open the discussion up to some itb/hybrid solutions... like perhaps 5042 on individual tracks or stems, that kind of thing.

Also, though, even with all the analog saturation in the world, there is still a separate issue on mixdown. Let's say an ssl is giving it some motion with a pumping effect. The first 10ms or so of each kick are untouched, and it dips after that... so unless one does ZERO processing of the master buss, all those kick attacks when placed along with the other elements will be causing the program to approach 0dbfs on each attack.

So, let's say we've done what we can about limiting and/or saturating the individual tracks, the question remains about how to then process the overall track.

Personally, I've been looking into a few concepts. The first is that if it's desirable for vibe and texture to get some sort of pump going, it may not be desirable for vocals. Therefore, it makes sense to me to do the ssl thing on everything BUT the vocals, and then to add the vocals on top of that. That way, they aren't pumping as well, and you can push the pumping a bit further if you wish for effect without undoing the crown jewels of any pop production (the vocals)

So... if you've already got some pumping via busses, already got some saturation and limiting on tracks and/or busses... then what do you do for mb?

[Apollo Soul]

Come to think of it... wouldn't a computer be better at the "Just divide all the amplitudes in half" thing?

There must be some reason things aren't done this way, right? Seems like it would be by far the most transparent type of compression. The only downside I can think of is noise floor at some point, but if you just had threshold control, and could set it for quietest bits, that would be fixed, no?

Is there a way to compress with no attack or release? Just straight ratio?

Would there be any reason NOT to do that? Seems it would be far less damaging than limiting.

[Apollo Soul]

Trying out a variety of things per track at the moment... from tape emulation to different comps, to apogee's soft clip.

In the meantime, who's got a take on the whole waveform simple ratio thing?

Where am I going wrong on this? How would that not be simple the most transparent possible way to increase density without changing anything else?

Is that essentially what some of those old Abbey Rd. compressors did with the "hold" feature?

Right... send a click through first or something so your first hit doesn't jump out... then it just compresses EVERYTHING EVENLY from there on out... with no attack, and no release.

Would this or would it not be ideal for dance music that is already mixed so it has no particularly prominant peaks?

If so, are there modern units that do this and/or is there a way to just do this part in software?

[Blast9]

Originally Posted by Apollo Soul

Right... send a click through first or something so your first hit doesn't jump out... then it just compresses EVERYTHING EVENLY from there on out... with no attack, and no release.

Would this or would it not be ideal for dance music that is already mixed so it has no particularly prominant peaks?

If so, are there modern units that do this and/or is there a way to just do this part in software?

ITB - This is where look-ahead compression/limiting comes in... Very useful!

I know of mastering engineers that use several comps/limiters in series with low ratios to get to that "transparent" density - Mind you, it is with expensive outboard gear rather than plugins... nevertheless, try it!

PS watch out with the Apogee soft lim - it is very obvious even on single channels if you hit it too hard - doesn't take much! In any case it is there to prevent accidental overs on the way in.

PS have you tried multiband comp? - I think Lagerfeldt sometimes uses it to pin down the low end

[hive]

Originally Posted by Apollo Soul

Someone had suggested printing the mix to analog tape. Unfortunately, that's the ultimate workflow killer since you don't get to hear your mix back until after the tape... unless you've got a setup where you can monitor off a playhead after the record head. It's been a ling time since I've used tape, and when I did, it was used for tracking, not this way, so I don't remember if that was doable.

it's simple. insert the tape machine via I/O plugin or whatever, monitor/print off the repro head playback.

no one's gonna be able to tell you whether that's the answer for you. you really gotta spend some time experimenting with it.

Originally Posted by Apollo Soul

So... first of all, I'd like to open the discussion up to some itb/hybrid solutions... like perhaps 5042 on individual tracks or stems, that kind of thing.

how deep do you wanna get? personally when writing in the box, I put Sonimus Satson on all my channels and saturate them to taste. then when I'm working on the final mix, I'll make decisions to leave plugins on or replace them with hardware inserts.

Originally Posted by Apollo Soul

Personally, I've been looking into a few concepts. The first is that if it's desirable for vibe and texture to get some sort of pump going, it may not be desirable for vocals. Therefore, it makes sense to me to do the ssl thing on everything BUT the vocals, and then to add the vocals on top of that. That way, they aren't pumping as well, and you can push the pumping a bit further if you wish for effect without undoing the crown jewels of any pop production (the vocals)

that's not unusual. create buss groups and process them separately, then leave your master untouched.

Originally Posted by Apollo Soul

So... if you've already got some pumping via busses, already got some saturation and limiting on tracks and/or busses... then what do you do for mb?

is sending your stuff off to a professional mastering house out of the question? that's your best bet man. get the mix how you like it and let them make it loud.

if not, then here are some ideas for you to do it yourself on the cheap...

push the mix into Airwindows' 3D Clip to knock down transients on the master but not so hard that your induce distortion. at that point you can probably get away with putting a limiter on after doing a db (or less) of limiting and not hate it. or maybe no limiter at all, really depends on the track.

try it and report back.

[hive]

Originally Posted by Apollo Soul

Trying out a variety of things per track at the moment... from tape emulation to different comps, to apogee's soft clip.

do you have a clean way to gain the mix up into the Apogees in the analog realm? if so, turn softclip OFF and experiment with clipping the ADC. a/b that against 3D Clip and see what you prefer.

[yotonic]

Merry Christmas to all.

[SCDBM]

This what your looking for? "Big" sound wise?

http://soundcloud.com/dropboxmastering/raphie-grooves-the-rush

[hive]

Originally Posted by Apollo Soul

I would put it this way. If you KNOW FOR A FACT before you even begin mixing a song that it will eventually be crushed to a certain degree, WHEN would be the right time to start acting accordingly?

Is it after the mix is done and you're sitting home waiting for the track to come back from mastering with your fingers crossed?

no. figure out a mixbuss master chain you like with your own plugins/hardware. listen with it engaged and bypassed while you mix so there are no surprises later.

aside from that, I'd recommend you work on processing stems before they hit the mixbuss..should get you where you need to go.

sorry if I'm missing something but that ominous wall of text is not friendly with my hangover

[Blast9]

Originally Posted by Apollo Soul

Any answers as to the pure ratio question? Is there a super clean way to just cut every amplitude in half?

Did you try lookahead comps/limiters on the busses or the master? Look into Flux plugins are transparent and super-tweakable.... as in "drive you nuts" tweakable but worth trying.

If you stick them on the busses - this would ultimately give you more flexibility with regards to controlling the subs on this track.

You'll have to delay any that don't have the plugin because of the look-ahead

[Apollo Soul]

Originally Posted by Blast9

Did you try lookahead comps/limiters on the busses or the master? Look into Flux plugins are transparent and super-tweakable.... as in "drive you nuts" tweakable but worth trying.

If you stick them on the busses - this would ultimately give you more flexibility with regards to controlling the subs on this track.

You'll have to delay any that don't have the plugin because of the look-ahead

I've tried flux before and used the lookahead... not tried to do endless release time, tho.

I don't need or want this on the busses, though.

The idea for the tracks and busses is to get all the limiting out of the way so all the errant peaks are fixed before the mixdown... and then to do any coloration or pumping desired (like how I have ssl on whole track without vocals... then vocals added in)

The no attack, no release thing would be for AFTER all that is done when the only peaks I have to deal with are the result of bass, guitar, kick, and snare all hitting on the same beat.

The idea is that if the tracks have all been limited well up unitl that point, then We're just dealing with no major peaks, but a sum of elements that puts it above the norm. In that case, I don't want to limit any further because we're past the point of peak limiting, and it will likely sound bad... but I could still do whole waveform compression, and effectively just reduce the dynamic range evenly overall... and that should still get more density, but without any more limiting.

Once each track and buss is limited, the only other way to keep those summed element peaks down would be to do something like using the kick to trigger a comp that drops the music down while the kick attack is happening. I haven't tried that on this track yet, but in theory, if done judiciously, it could do just enough to keep those moments where the kick attacks from stacking up quite so high... and get an extra couple db overall.

So, really, this last bit I'm looking for could be a mastering stage as the idea I'm trying to work with here is the one of peak limiting being something that should be done before the mixdown only... Ideally mastering would be limited to invisible compression and any other stuff like summing low end to mono, etc.... as opposed to the popular notion that you should ignore all this stuff, and send it off to a ME who is likely going to limit the crap out of it.

Realistically, I'm sure mastering will also involve a touch of digital brickwall... but the whole goal here is to minimize that need.

[Cheebs Goat]

Apollo, you need to take a step back and reevaluate this whole thing. Mix it exactly as you want it to be, then worry about the loud.

[Blast9]

I may be getting this wrong, but if you want to control "a sum of elements that puts it above the norm"...

...what you are describing sounds like levelling, i.e. low threshold, very fast attack / very slow release, but you'd likely need to:

1) use a very low ratio in order not to affect the punch/attack

2) The other option is to use volume automation if it is just a few peaks.

-

Also you would have done most of the work here anyway:

I have found it's much less damaging to deal with dynamics per track so that there are no such massive spikes on mixdown, and the MB compressor can reside comfortably in it's sweet spot.

Any levelling limiting/multiband comp after this stage will be gentle "peak management"

[Apollo Soul]

Originally Posted by Cheebs Goat

Apollo, you need to take a step back and reevaluate this whole thing. Mix it exactly as you want it to be, then worry about the loud.

I disagree.

Frankly, I find that this advice totally is out of place when dealing with music that will end up being near 0dbfs on each and every beat.

To do so would mean that the ME would have to do VERY destructive things to the entire mix... when those same peaks could much better and much more transparently have been handled on an individual basis before the mix.

Which is more destructive to the mix?

A: Tailoring an individual peak to drop by 10db where you can isolate it and have complete control over it.

B: Using a limiter on the whole mix to drop everything suddenly by 10db to deal with that same peak making large doses of digital brickwall all but unavoidable, and causing very audible artifacts in the other elements of the mix.


Sticking my head in the sand about the mastering process isn't doing me, the music, or the ME any favors.

The more I ignore this, the more the ME HAS to crush the whole mix.... and to do so in much more destructive ways.

I was hoping this thread might elicit a dialog for rational approaches that don't involve ignoring the issue and assuming the ME has magical powers.

And yes... I'm talking about very low ratio mastering with a threshold WAY down if not all the way down... and lookahead would solve the attack part.

Now, that being said... yes, anything done from mix on for the pure purpose of more density would be a mastering stage, and not a mixing one...

... just being conscious of the ability early on in the game to create a situation where digital brickwall limiting is NOT necessary during the mastering stage... whereas it absolutely would be necessary if I just ignored the issue and used my ears only rather than thinking ahead and dealing with the peaks on a per-track basis.

[Cheebs Goat]

So the idea here is dropping all pretense that anything is more important than the final volume.

In that case, yes you probably will get hella' loud results with this. And yes, you will get something that does not get altered during mastering. You will also get something that is unusually limp.

The "traditional" method creates transients that stick way out of a mix and then get crushed with compression, limiting, or clipping on the mix bus. The entire production reacts to that process. Everything sucks down, farts in unison, and then releases. It gives a clue that there was once an impact there. It does it's best impact impression.

If that transient is flattened at the track level, nothing happens to the mix once it "hits". It's just a noise that buzzes on by without ever making it's presence felt on the other tracks or the listener. The song isn't even doing an impression of an impact anymore.

That horrible 2buss distortion and artifacting is pretty much the only tool left to impact a listener with anymore. Remove it at your own risk. Only constant buzzy noise will remain.

[daez]

I would handle the occasional peak with a fader ride or audio edit as apposed to a comp or limiting. If you want a smaller gap between your peaks and rms it would be better to mix it that way but that goes back to what cheebs goat said. It's easy to take the peaks up to 0dbfs with out ruining the mix but once your peaks hit 0 if your rms is still unsatisfactory that is a much bigger mountain to climb unscaved. Good luck and hope you find what your looking for...

[Apollo Soul]

Originally Posted by Cheebs Goat

So the idea here is dropping all pretense that anything is more important than the final volume.

Not in the least.

Franky, where I'm coming from is pretty far on the opposite end of the spectrum... at least as far as top40 dance/pop produced in the last couple of decades could be.



Stated goals:

1 To create dance/pop that is not pushing the loudness envelope, but in a generally competitive loudness range with modern productions while maintaining greater sonic integrity than most of the comparable tracks

2 To do so while minimizing if not eliminating the need for the most destructive thing of all: The use of extreme limiting on the master.



I really don't think it's that hard to understand. If the individual tracks and stems/busses are tamed and thickened before mixdown, then there will still be peaks... they just won't be so wildly varied.

I still WANT peaks. Like I said... I WANT to get an ssl style pump going for the purposeful introduction of motion that it gives.

The problem is that something like an ssl isn't touching the first 10ms or so of the signal... so the question remains... what do you do about those peaks?



Well, as was mentioned... if those peaks are more consistent, then half the density battle is already won as the distance between peaks and rms is lessened.

If not... the WORST possible time to deal with it would be during mastering.

That extra 5db jump that's above and beyond the next downbeat isn't going to lead to a satisfying pump from a compressor, it's going to be pure distortion from the ME's most aggressive limiter. The pump from the compressor is already there... just like it would have been if the peak had been tamed a bit first. The ssl reacted exactly the same to it either way. The only difference is what has to happen to the mix later if it isn't tamed first.


Once the egg is scrambled, there's really no way to tame errant peaks WITHOUT destroying the mix.

[hive]

why are my RMS levels still so low?

[Lagerfeldt]

Originally Posted by Blast9

I know of mastering engineers that use several comps/limiters in series with low ratios to get to that "transparent" density - Mind you, it is with expensive outboard gear rather than plugins... nevertheless, try it!

PS watch out with the Apogee soft lim - it is very obvious even on single channels if you hit it too hard - doesn't take much! In any case it is there to prevent accidental overs on the way in.

PS have you tried multiband comp? - I think Lagerfeldt sometimes uses it to pin down the low end

Serial or parallel compression are useful techniques - both can be quite transparent when used correctly. Digital parallel compression can be extremely transparent since it's phase aligned and by design doesn't change the dynamics of the original signal.

Regarding multiband comp: it can be very useful for taming the subs. However, downward expansion may be even more useful if you want to tighten up the lows. Or a combination of both.

Generally I use multiband only if regular equalizing and compression won't do the trick. I prefer the Flux Alchemist which is a minimum phase dynamic processor capable of both compression and expansion, including true upward compression.

I tend to do serial compression only in the analog domain while I do parallel compression both digitally and with outboard, depending on what the purpose is.

[Apollo Soul]

OK, well I have a sample of a track in my signature now.

I did many experiments. It may not be the best example for dance music since the drums are on the soft side, so I was able to push things a bit harder and mess with the transients more than in some forms of dance music.

Basically, what ended up sounding the best was going in per track and using comps, limiters, and extensive automation to even things out. As fot the master buss, I ended up going with:


1 The Glue doing 2.5 max GR ssl style pump... so the peaks are still untouched as it doesn't kick in for 30ms. Actually, this is on the "karaoke" buss which then has the vocals added to it for the MB... just helps keep the vocals from ducking around too much.

2 Master buss has Slate fg-x buss comp (limiter disabled) This is doing max gr of 1db at 1.1:1. Very mild, but I've found it helps to impart a smoothing quality... especially with all the high end tambourines, bottles, etc. Again, peaks untouched as it doesn't kick in for 10ms.

3 G10. Usually, I wouldn't put more than -2db of gr using this as that seems to be the magic number, but I pushed it to -3 in this case since there was a retro sort of vibe going on sonically, and it seemed to want to be dirtied up a bit. Key here really is the tube makeup gain which is all the way up. Again, peaks untouched although I don't know the exact time to kick in (attack around 2 oclock, release around 10 oclock.)

4 Apogee Duet 2 soft clip. I did MANY experiments with this and yes, it can ruin a mix. I've also found that depending on the type of drums, if you set it so the absolute highest peak just barely triggers the over light, and then engage, it will just barely start to round things off without ever really hacking at them... overall, the level drops the highest peaks by around 1 db.

5 More peak work with The Glue's softpeak feature (REALLY impressed with this one) This will redraw the waveform for the final 2db of peaks. So this is completely reshaping the attack of all the kicks and whatever hits simultaneously. I tried a dozen tools for the task, and it blows them all away. I also played around and found that if I engage the compressor as well with the threshold all the way up, since I'm so close to 0dbfs, it will just barely tickle the meter. Dropping the attack to 10ms, and giving it a fast release just gives a very subtle emphasis on the rhythm as the attacks pop out a bit. Between this and the softclip, seems to really be what the doctor ordered. They're emphasised, yet beaten into submission.

6 Event Horizon. This one I have mixed feelings about. You definitely HEAR what it's doing... and not often in a good way. Again, though, retro vibe, dirty sonics and all, so it seemed to work. VERY subtle 0.5db of limiting.... which seemed to also really emphasise the bass and do more of a track thickening than anything else. Dialed it way down, but track just seemed thinner without it. Try it. Certainly unlike other digital limiters. It's not just shaving peaks.

7 Branworx control. Very mixed feelings about this one too. Mainly, I just wanted to put the low end in mono... ended up going with 150hz or so. I did end up using a touch of stereo widening, though, which I hate the idea of, but the image had narrowed at this point. Triple checked in mono, etc... and it seemed to be just a touch better with 110% wide setting.

8 Another instance of the Glue doing softclip with same settings. Have I mentioned how much I love this thing? So I've really just flat out redefined the peaks at this point... don't know that I would've pushed this so far on something more like house music.

9 Elephant. The whole goal here was to AVOID digital brickwall limiting, so it's mainly just grabbing a hair here and there where one channel is hotter than the other. Overall, I think the absolute highest thing it grabbed was 0.7db... and then used to set output ceiling at 0.3db to avoid intersample peaks.


So... hear the results for yourself, and comments welcome, but here's my take... I still need something else, but that something else is NOT more peak limiting. If anything, I've carried that bit too far and done too much damage already.

I feel like what it needs is for the entire waveform to be squeezed... but without more pumping, without more emphasis on peaks, etc... so basically something like a hold.

As usual, Lagerfeldt is making me look up terms (should have remembered them from when I went through this for vocal compression).

Questions, comments, refreshments will be served in the lobby after the q and a.

edit. ok, seriously, what's the deal? I cannot figure out why my sig shows up sometimes and not others. This is not a short post. The signature does not appear anywhere else on this page... why isn't it showing?

[squirreltrench]

Originally Posted by Lagerfeldt

Digital parallel compression can be extremely transparent since it's phase aligned and by design doesn't change the dynamics of the original signal.

+1... digital parallel compression can be truly amazing. You can read more about it in Katz' book on mastering. Threshold = as low as you can go, like -50db. Ratio - depends on your track of course, but never greater than 2.5:1 and more like 1.4:1. Then turn up your parallel comp'd track back into the original and like magic, it's louder with no or little nasty "traditional" compression artifacting. Attack = 0ms or maybe 1 or 2 ms, and I find a release of about 8-10ms does the trick.

Used in small doses, it is pretty much completely transparent. You can hit this technique harder (by putting more of the compressed track into the original), and a trained ear can hear the loss of "airyness" of the track, but it does get it louder without the squash-tails associated with just putting something through a limiter.

[Apollo Soul]

I've been doing quite a bit of it actually.

I've been trying infinite release as well.

Also, since this thread was active, I've tested out a couple of ssl clones and...

shoot me, but I PREFER the Glue.

Also been digging into some more articulate final limiting options. As an experiment, I tried to see how far I could dial in a brickwall. Current track in siggy is crushed down to almost -6db. It does distort a bit, but it's not too bad. Still working on how to do less of that, and more of the other stuff.

Running some individual tracks through tubes, trannies, and whatnot has helped, although that wasn't done on this track.

[Arksun]

Originally Posted by Apollo Soul

Which is more destructive to the mix?

A: Tailoring an individual peak to drop by 10db where you can isolate it and have complete control over it.

B: Using a limiter on the whole mix to drop everything suddenly by 10db to deal with that same peak making large doses of digital brickwall all but unavoidable, and causing very audible artifacts in the other elements of the mix.

In both cases you're effectively talking about limiting though, because to have as dramatic an impact to truly squash that 10db peak on an individual channel you'll need a limiter, or an extremely fast attack compressor, but at those speeds you increase distortion anyways.

So you could almost look at it as limiting whole mix vrs limiting each individual channel.

I'm with Cheebs Goat on this one. I would focus on just getting as high quality dynamic interesting punchy dance mix as you possibly can first and foremost. The best engineered mixes are the ones that survive the limiting better at the final stage imho, and by best engineered I mean very well balanced, not necessarily compressed to within an inch of its life. Sure, it may mean the limiter seems to do less damage, but was it really worth it for all the sonic compromises made at the mixing stage?.

A couple more things being an ITB EDM person that you probably knew already but are handy tips:

1) Not all soft-synths are in phase with each other, but generate the waveform inverse. Flipping phase on its channel can help improve things particuarly with bass/kick interaction.

2) Some soft-synths can generate large amounts of sub-20 DC offset type energy which sucks up a lot of bandwidth, this would need shelfing off with eq.

[Apollo Soul]

Originally Posted by Arksun

In both cases you're effectively talking about limiting though, because to have as dramatic an impact to truly squash that 10db peak on an individual channel you'll need a limiter, or an extremely fast attack compressor, but at those speeds you increase distortion anyways.

So you could almost look at it as limiting whole mix vrs limiting each individual channel.

I'm with Cheebs Goat on this one. I would focus on just getting as high quality dynamic interesting punchy dance mix as you possibly can first and foremost. The best engineered mixes are the ones that survive the limiting better at the final stage imho, and by best engineered I mean very well balanced, not necessarily compressed to within an inch of its life. Sure, it may mean the limiter seems to do less damage, but was it really worth it for all the sonic compromises made at the mixing stage?.

A couple more things being an ITB EDM person that you probably knew already but are handy tips:

1) Not all soft-synths are in phase with each other, but generate the waveform inverse. Flipping phase on its channel can help improve things particuarly with bass/kick interaction.

2) Some soft-synths can generate large amounts of sub-20 DC offset type energy which sucks up a lot of bandwidth, this would need shelfing off with eq.

Yeah... it's assumed that if you want to have a huge impact on the peaks one way or the other, there's some sort of limiting.

I've been doing LOTS of experimentation, and I'll have to respectfully disagree philosophically... at least for the style of music I'm doing. It really does seem to help to pin things neatly before mixdown, and it does quite clearly lead to less distortion for the same RMS come mastering time... which is really the ultimate goal here.

Since starting this thread, I've redone three mixes now with both approaches, and reached the same conclusion each time... treating the dynamics on a track by track basis leads to less distortion in the mastered track at the same rms levels. Thickening them in various ways can help as well... but yes... limiting individually as well where necessary.

I flip phase on the kick sometimes too... depending on what the attack portion looks like.

One thing I'm trying to nail right now, is the right formula for dealing with moments where the kick and bass hit simultaneously. That's the bit that's going to distort (and distort any vocals or other instruments with it) first during mastering, so working on variations of ducking bass slightly during kick attack, etc.

Track in siggy is good example. RELATIVELY distortion free despite some heavy duty limiting (more than I usually do), but when the 808 and kit kick hit simultaneously when rap starts, it does distort... despite dropping kick volume, eq'ing low end out of kick, etc.

Still haven't got that part down.

edit... oh, and yeah, I HPF like crazy... and usually well above 20hz. If I can't hear it on a typical car stereo, then it's just robbing headroom IMO.

[Apollo Soul]

One thing I'm just not understanding...

I've been doing quite a bit of experimenting with using low ratio, full waveform compression (not sure if this is what's being referred to as upward compression?)

Anyway, using something like Waves Renaissance, or similar with shortest possible attack, and a ratio of 1.25:1, when the threshold is all the way down, it shows a GR of around 6db.

I've done some variation of this many times, but I keep noticing the same thing:

If I take the compressed and non-compressed version, and alter the output gain of the compressor until both versions have the same identical peak value...

... then run both through the final limiter, the UNcompressed version has a higher RMS.

This should not be the case in theory, but I've replicated this a dozen times now. Unless something's wrong with the metering, I must be missing something... or is it that it doesn't have a true 0ms attack... so occasional peak gets through, and rest gets compressed...

... so to make this work, I'd need a 0ms attack, right? So what comp does that right? I'm seeing lookahead on lots of limiters, but not low ratio comps.

[Apollo Soul]

So... many experiments later...

I'm finding that when producing music that is both dance AND pop, there is a bit of a bugaboo with the vocals. They tend to fare worse for heavy limiting, and bob and weave badly when there are breaks in the song, etc.

It seems better to do a "karaoke" mix, and then lay the vocals in so if there's any bits with a sudden heavy hit to them, the vocals don't get a weird flutter in them at that moment.

If you process the music first and add vocals, though, still wondering what the best way is to combine them. Should the vocals be used as a sidechain to limit the music a bit (mild ducking?) No compression on overall mix, but limit a bit?

Anyone else doing the compression and limiting of the music first and THEN adding the vocals? If so, how are you combining and processing from there?

[Sudad G]

a lot of old school House producers still keep those Mackies around to do just that, I know.. It's that analog clipping on drums that makes old house sound so thumpy and sexy.

That's exactly what I do and what I recommand still today for that kind of music.

I saturate the drum and bass sounds (with my Mackie Analog Mixing Console) and compress single tracks very heavy (sometimes ratio 8:1 and a gain reduction -10dB) with my DBX Compressors and after the stereo bus or sub groups just a little bit with the t.c. finalizer that I also use for the A/D conversion into the DAW. After recording the compressed and saturated groups or single tracks into the DAW I add hard- and software-effects on it and compress them again with just a little bit of the UAD LA-2A. Finally at the end I limit the complete master bus with Waves L2 with a gain reduction only between -1 or -2dB.

The vocals I process only in the DAW. At first with the UAD 1176 and second UAD LA-2A. The UAD 1176 controls very effective and fast the volume of the vocals and the LA-2A makes the aggressiv sounding vocals (from the hard compression of 1176) sounding very smooth and softer. To get the vocals placed in the mix (playback) I use the hardware reverb processor Lexicon 300 (preset Contemplate or Dance Hall), which is awsome for that work. I use it in Cubase 6 as an external plugin. It's difficult to place vocals into the mix with native plugin reverbs. With hardware like Lex 300 it goes from alone and this saves time.

Thats's it.

Sudad G