The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Compression: "Common Knowledge" Questions Dynamics Processors (HW)
Old 13th August 2008
  #1
Lives for gear
 

Compression: "Common Knowledge" Questions

A) I was told a long time ago that it's best to strap a compressor over a channel before any EQ, because the Compressor would tend to flatten out/work against the EQ dips and peaks.

Is this true?

Are there any other "common sense" Compressor related things that novices might benefit from knowing?

B) Does Lookahead mess with Attack time? More Specifically, does lookahead act like negative attack? As in, keeping your attack at 0, and with a lookahead of 30ms, will the compressor start clamping down 30ms before peak arrives at the playhead?
Old 13th August 2008
  #2
Lives for gear
 
Unclenny's Avatar
I like EQ first....lets the comp work with the sound I have in mind.
Old 13th August 2008
  #3
Lives for gear
 

Thinking about this again... Does the Compressor even out all frequencies?

I mean, EQ is like a compounded/vertical gain control, so at a given moment, you turn down the gain of one frequency while boosting another..

A compressor is more horizontal gain control, changing gain over time...

So, my first statement is flawed in that the compressor won't flatten out the EQ changes because at a given moment, the compressor will reduce the gain of the OVERALL sound all at the same time, so all of your EQ curves stay intact, just lowered in volume.

Glad I posted this, cause I think I've been misinformed =)
Old 13th August 2008
  #4


Frequency is related more to the attack and release than the ratio.

EQ after the compressor can be the right way to go too, especially if it's set up for a quick attack and starts sounding a bit thin.



-tINY

Old 13th August 2008
  #5
Gear Addict
 
macrae11's Avatar
 

No you're not completely misinformed. If you have a significant EQ boost, it will trigger the compresser earlier. Yes once it's triggered, it will compress all frequencies, but it does change how the compresser works.

Doing a cut, will not have as quite much affect on the EQ. However if you are reducing a louder part of the signal significantly, like the low end of a bass guitar, it will affect the threshold of the compressor noticably.

There's no real correct way do things. You just have to decide on a case by case basis what sounds best for you.
Old 13th August 2008
  #6
Lives for gear
 
Unclenny's Avatar
Quote:
Originally Posted by tINY View Post
EQ after the compressor can be the right way to go too
Hmmmm........

Cut offending freqs, compress........boost to taste.
Old 13th August 2008
  #7
Any EQ move will indeed effect how compression will react.

EQ is tonal adjustment tool, so sometimes its necessary to tonally affect something before you compress it and make it louder and less Dynamic. The Reverse is certainly true as well, as it will often be necessary to EQ something POST compression, because you DONT want any EQ move to affect how the compression goes down [the same can be said with VOLUME AUTOMATION and such processing].

I have often found that compression can be very useful as a tonal adjustment device, as you are manipulating the TIME DOMAIN with attack [onslaught of compression] and release [when the compression cell "lets go"] This can GREATLY help frequency response of a source in many ways. BUT flipped around, don't ever ask an equalizer to act like a compressor, cause unless where talking about multi-band or frequency dependent compression [De-Essing, multi-band compression...etc....] than its another thread.

The Thing is BOTH methods are means to an end. Never let anyone tell you something is correct or incorrect with these kinds of questions. There are no correct answers, but the idea is for YOU to come to the correct equation by figuring out what works for you, and I think ultimately these are preferences and technique, but a god engineer will know when to use the proper tools and techniques, when he/she needs them.
Old 13th August 2008
  #8
Moderator
 
TonyBelmont's Avatar
 

Quote:
Originally Posted by DarkEcho View Post
A) I was told a long time ago that it's best to strap a compressor over a channel before any EQ, because the Compressor would tend to flatten out/work against the EQ dips and peaks.

Is this true?
Short answer is Yes. If you boost a frequency, it will affect the compression, and the manner the compressor behaves (often compressing more at said frequency). The same can be said for cuts. It's up to you to determine which order you want for any given situation.
Old 13th August 2008
  #9
I don't even try to think about it intellectually any more. I just try it both ways. One will sound better than the other, take that one. Sometimes it'll sound better letting the compressor work harder on the un-cut signal and then EQ that. Sometimes it'll work better letting the compressor work on the alreayd cut signal.
Old 13th August 2008
  #10
Lives for gear
 

Quote:
Originally Posted by TonyBelmont View Post
(often compressing more at said frequency)
Wait, I don't get how this is possible.. If a compressor is reducing the gain of a track, its not going to reduce the gain of a specific frequency of that track, it will reduce all of the frequencies at the same time, right?

A compressor is going to affect all frequencies (as one whole), all the time.
Old 13th August 2008
  #11
Gear Addict
 

I do it both ways, but more often I will have the compressor post EQ. If you have a plug with both and will allow you to switch the order of processing, play with it and you'll see how it affects the sound
Old 13th August 2008
  #12
Lives for gear
 
FreshSkweez's Avatar
 

One engineer showed me a way of approaching post-EQ compression that changed the way I deal with percussive instruments forever.

Say you take a kick drum sound and accentuate the attack with your EQ (say 100 Hz and 4K), then you strap a fast comp after the EQ and make both frq bands hit it at the same time. The result is a super tight attack that'll cut through the mix like crazy.

In my personal workflow if you break it down most sounds go through:
Corrective EQ - Compression - Creative EQ - Limiting (or gentle transient compression)
Old 13th August 2008
  #13
Moderator
 
TonyBelmont's Avatar
 

Quote:
Originally Posted by DarkEcho View Post
Wait, I don't get how this is possible.. If a compressor is reducing the gain of a track, its not going to reduce the gain of a specific frequency of that track, it will reduce all of the frequencies at the same time, right?

A compressor is going to affect all frequencies (as one whole), all the time.
You are wrong.. you are making the compressor work more at that specific frequency range when boosting.

Take a track and boost 1khz by 15 db and then compress it and you will understand the concept.
Old 13th August 2008
  #14
Quote:
Originally Posted by DarkEcho View Post
Wait, I don't get how this is possible.. If a compressor is reducing the gain of a track, its not going to reduce the gain of a specific frequency of that track, it will reduce all of the frequencies at the same time, right?
UH oH......You opened up the can!

Multi-BAND compression!

It is possible to create frequency dependent compression by manipulating the compression circuit. Most often [in our engineering world of using the gear and not actually building it] this can be achieved with an equalizer patched into the detector circuit, [which can be called a Side-Chain] so as to make the compressor compress, only the frequency range you want [given a specific threshold] and discard the frequencies that you don't, through the detection element.

If your equipment has this possibility then you simply copy the signal with a MULT on the patch bay and send one half to the compressor's input and the other half to an equalizer which will be passed into the SC of that compressors input, and then you BOOST all the stuff you want to TRIGGER compression, and CUT all the stuff you DONT. The point is the compressor jumps to compress the 8k bump you send into the SC, determined by the TIME you set on the compressor [onslaught and release] and intensity of effect depending on the threshold.

There are many compressors that feature HP filtering in the detector as a standard, and you might find this feature on specific "Buss Compressors" as they're design is geared towards this problem, but its also a problem addressed with many other designs as well. For 2-buss work, a targeted filter is a very important feature to have.

A few examples of compressors that feature switchable or patchable detector circuit filters/SC's would be the Drawmer 1968 ["BIG"-switchable HPF detector], the Dramastic Obsidian [switchable 150HZ HPF detector], the Crane Song STC-8 [patchable side-chain], the Manely ELOP [switchable HPF detector, 100HZ, 200HZ, FLAT], the Thermionic Phoenix MV [switchable HPF detector..], FMR RNC/RNLA [patchable side-chain] Some of the best compressors in history are built from this principle [with regard to the "side-chain" and detection or compression cell] There is some crazy **** going on in a detector circuit!!

If you ask me what the best compressor is, I am going to tell you the InnerTUBE Atomic SqueezeBox, and the GML 8900 [but they don't even call that a compressor, its a controller!] because of what they are capable of sonically with regard to the action of reduction. The Squeezebox is not a limiter nor compressor, actually....we're not really sure what it is, but all I can tell you that its most certainly a "SqueezeBox" that doesn't squeeze the tone....I cannot figure out if I like the amplifier or the gain cell better.
Old 13th August 2008
  #15
Lives for gear
Quote:
Originally Posted by DarkEcho View Post
A) I was told a long time ago that it's best to strap a compressor over a channel before any EQ, because the Compressor would tend to flatten out/work against the EQ dips and peaks.

Is this true?

Are there any other "common sense" Compressor related things that novices might benefit from knowing?

B) Does Lookahead mess with Attack time? More Specifically, does lookahead act like negative attack? As in, keeping your attack at 0, and with a lookahead of 30ms, will the compressor start clamping down 30ms before peak arrives at the playhead?

If you're in the mixing phase (all tracking complete), then I find what works best for me is to try and get a decent sounding mix first just using faders and panning. If any particular track sounds weak in the mix, simply try to turn the fader up and see what happens before reaching for a comp. or EQ. If it's sounding better but masking other stuff now, then find the problem frequency and cut some of it out. I usually reach for comp. last after getting the whole mix to sound good (and balanced) without any at all. Then you will find that each comp. you do add afterwards will only enhance the mix further in the direction you were going. You have to be patient, because the mix is not going to sound incredible right away. I will often end up with as many as 20 outboard channels of comp. in a mix (some inserts, some paralell); but it usually after I have done all the EQing that needs to be done.

I actually learned this method from a famous producer/engineer who also had his mentor teach him the same thing. Their philosophy was that if you can't get the mix to sound "decent" and spacious without any comp. at all, then reaching for comps. is just going to compound the problems that are already there. Also learning how to use EQ for cutting out frequencies is very important. Those who always reach for the boost knob are usually just creating more problems as well.
Old 13th August 2008
  #16
Lives for gear
 

Quote:
Originally Posted by TonyBelmont View Post
You are wrong.. you are making the compressor work more at that specific frequency range when boosting.

Take a track and boost 1khz by 15 db and then compress it and you will understand the concept.
The way I am imagining this is that if I boost a freq by 15db, then strap a compressor with 15 db of reduction, that it will UNDO the EQ and it will sound like I'd never done the EQ at all.

That doesn't make sense to me, but i'll try it.

I'd think that the comp would turn the whole track down, but 1khz would still be 15db louder than the unboosted frequencies. No?
Old 13th August 2008
  #17
Gear Nut
 

DarkEcho, you're right that a compressor works on the signal as a whole, try thinking about it this way.

If you boost 15dB @ 1KHz, there is a 15dB difference between the stuff at 1KHz and the rest of the signal. Now think about what a compressor does, it makes the loud stuff quieter with gain reduction, and then makes quieter stuff louder with makeup gain. This means the compressor will take the boost at 1KHz (the louder stuff) and turn it down, then it will take the quieter stuff (the original signal) and bring it up, so those two signals are now closer together. This is what people are talking about when they say the compressor flattens out your EQ. It works the same for a cut, when you get rid of certain frequencies the compressor will bring them back up when you use the makeup gain.

Compressors behave this way with stuff besides EQ too, if you think about something with a loud, low pitch, fundamental and then quieter higher frequency overtones, a compressor will bring down the level of the fundamental and then bring up the level of the overtones. If you look at it from an EQ perspective it's cutting the fundamental and boosting the overtones, that's how compressors effect EQ.

With this in mind, use the order of compression and EQ whatever way works best for you. Just know the more you compress something, the more it's going to flatten out your EQ, so the more you should consider experimenting with putting the EQ post compression.
Old 13th August 2008
  #18
Lives for gear
 

Ok, maybe it matter what the signal is..

lets say there are two sustained tones in the track, a 1k and 2k tone

both are at 0.0dbfs

You boost 1k by 3dbfs, so you have 1k at 3db louder than 2k.

Then you apply a compressor after that, with 3dbfs reduction acting on the signal.

The 1k will be dropped down to 0.0dbfs, and the 2k will be dropped to -3.0dbfs

I don't see how, in this specific situation, the compressor could possibly be flattening the EQ out at a given moment.. it turns down everything during one instance.


NOW!

Maybe we are both in agreenment but arguing about something else?

For istance, if you have a simple Hat-Hat-Snr-Hat:| drum beat and the snare is quieter than the hat, and you compress it, The compressor will turn down the whole track when the hat hits the threshold, but may release the volume crunch in time for the snare to hit, making the louder hat and quieter snare sound closer to the same volume, right?
is that what you mean?

Of course, if the snare and the hat hit at exactly the same time, the Hat will trigger the compressor and the compressor will decrease the volume of the Hat AND the Snare by the same amount.
Old 13th August 2008
  #19
I think that part of the confusion above is that, if you boosted 1K 10dB on a given instrument's track, it's only going to give maximum boost when there's a primary tone at or close to the 1K frequency.

So if you are doing this to a single line lead track, as you come close to that narrow boost, the compressor will clamp down on those notes a lot, but you won't necessarily be getting that much boost as you play notes far from the boost. You'll still have harmonics there and some boost, but not a constant 10cB boost. So the result will be that the compressor will just clamp down on the boosted frequencies and since there's only one note at a time, it'll not affect anything else.

If you did the same on a stereo mix down where there was always strong content in that boosted frequency range, the compressor would always be working hard because of that boost, and would pump the rest of the content and push it down even though it's not boosted.

I.e. the results will be content dependent, right?
Old 13th August 2008
  #20
Lives for gear
 

I think you are right. This is what I meant by having a snare and a hihat hit at the same time versus in a 1-2-1-2 fashion. If both the snare and the loud hihat hit at the same time, the compressor is going to take them both down equally, but if the loud hihat and snare are staggered, depending on attack/release, the snare may be completely unaffected, thus "flattening" out the track/frequencies (since it will only be acting on the loud/boosted hihat frequencies)
Old 13th August 2008
  #21
Gear Nut
 

Compressors don't just turn loud things down, they also turn quiet things up (that's the makeup gain part). If you use an EQ to make certain frequencies of a signal louder than others, then send it to a compressor, the compressor will reduce the volume of the loud stuff and bring up the volume of the quiet stuff. This means whatever EQ curve you had before shows up flatter, since the boosts are reduced and the cuts are amplified.

Think about it like a limiter on your master fader. You turn up the volume until it starts chopping off the peaks of the waveforms, which means you just made all the source material louder, and you made the peak information quieter. If you think about the peak information like your EQ boosts and your source information as your EQ cuts, you can see how the limiter has made the dynamic range smaller. If the peaks used to be 10dB above the source level, maybe now they are 5dB above the source level. If your EQ used to have 10dB dynamic range from the highest boost to the lowest cut, it now has 5 dB of dynamic range between those points.
Old 13th August 2008
  #22
Quote:
Compressors don't just turn loud things down, they also turn quiet things up (that's the makeup gain part).
But that assumes you use the makeup gain, which isn't a requirement. Makeup gain is a feature that most compressors would have, but it's not part of compression proper really. I pretty regularly don't use any makeup gain if the signal is hotter than is needed in the end. I'll just let the compressor eat up that excess.

The other thing is, they often don't reduce peaks anyway. They only do that if you have the attack short enough. They can often increase the effective dynamic range by leave peaks in place and pulling down on the subsequent inter-peak signal.
Old 14th August 2008
  #23
Lives for gear
 
Unclenny's Avatar
Excellent discussion!

I also refrain from makeup gain (not always) and boost the gain on the track after the comp has done its work.

Attack and release settings are SO crucial to getting it right.

Learning to compress musically takes time and practice.


Sugarland Rain
Old 14th August 2008
  #24
Lives for gear
 
synthoid's Avatar
 

Makeup gain is a red herring in this case. The quieter material becomes louder relative to the louder material regardless of makeup gain. The makeup gain simply determines the overall volume level of the output signal.

-synthoid
Old 14th August 2008
  #25
Here for the gear
 

Dean Roddey has it right, but it seems like some people are confusing time and frequency domain concepts and describing eq flatting by compression as if the compressor was multi-band. This definitely happens, but in the case of a fullband compressor it is entirely material dependent like Dean said.

Compression, in general, works in the time domain and not in the frequency domain. It won't flatten out your eq the way most people seem to be thinking... yes compression will turn down peaks and make quieter parts of the signal seem relatively louder than they did before if you use gain after compression, but these "louder" and "quieter" parts of the signal are in the time domain, NOT the frequency domain. If you boost 15dB at 1kHz, a compressor will not decrease the amplitude of the 1kHz frequency component of the signal but leave the rest of the frequencies alone to be turned up with make-up gain. This is what it seems most people are agreeing on here... that would only happen in a multi-band compressor where the band containing the 1kHz peak would likely be triggering compression much more often than the rest of the bands, given thresholds are set the same for all of the bands.

What happens when you eq before compression is simply that you change the threshold of the compressor, which someone has mentioned previously. By boosting the signal by 15dB at 1kHz, you will have increased the overall amplitude of the signal in the time domain since frequency components all add together to generate the time domain signal. Thus, if there is significant frequency content in the signal (if you are not boosting from like 0 to 15dB at 1kHz, and that part of the signal actually is a notable part of the sound in question), boosting by 15dB at 1kHz will increase the overall amplitude of the signal by a certain amount, proportional to how much of the energy of the sound is being boosted. The overall result is that since the signal is generally louder, it will generally trigger compression more often, but when that happens the entire signal will be turned down, NOT just the part of the signal which corresponds to the boosted frequencies around 1kHz. With makeup gain, quieter parts of the signal ( quieter in the time domain) will seem louder in comparison as usual. Note though that this process won't really "flatten out the eq" you applied pre-compression though.

If for some reason the frequency content of the track changes drastically to the point where maybe sometimes a the frequency content around 1kHz basically makes up the sound and sometimes there is no frequency content (pre-eq) around 1kHz, this will have the effect of turning up the sections where the signal is mainly contained in frequencies around 1kHz via eq, but then triggering compression on those segments as the signal goes over the compressor's threshold. Sections with a lot of frequency content in frequencies other than 1kHz would not see as drastic an increase in total amplitude after eqing and would basically be left alone and only compressed if they were above the compressor's threshold anyways. I don't know if this is clear at all, but basically what I am trying to point out is that the only way for a fullband compressor to affect eq is if the frequency content of the signal is actually changing pretty drastically which may or may not happen depending on the content you are dealing with. I feel like with most acoustic instruments, this wouldn't really happen unless you do some kind of drastic, surgical eq on it to create a big enough difference in amplitude depending on what notes are being played. But why would you really want to do that anyways?

I could see it maybe coming up with some electronic sounds / instruments though since they can have strange spectral makeup sometimes.
Old 14th August 2008
  #26
Quote:
Originally Posted by synthoid View Post
Makeup gain is a red herring in this case. The quieter material becomes louder relative to the louder material regardless of makeup gain. The makeup gain simply determines the overall volume level of the output signal.

-synthoid


Actually, no. The compression part has to do with lowering gain when the input gets too hot. Without some sort of make-up gain, the tracks getting compressed only get less loud.



-tINY

Old 14th August 2008
  #27
Lives for gear
 

Exactly.

If a loud sound happens at the same time as a 10db quieter sound, the compressor will turn them both down equally, and the quieter sound will still be 10db quieter than the loud sound, but they will both be quieter than they originally were.

Meaning, if the gain reduction is -10dbfs, the quiet sound (originally -10dbfs) will end up being -20dbfs and the loud sound (originally 0dbfs) will be -10dbfs.

However, if the loud sound preceeds the quiet sound, depending on the time and the comp. settings, the loud sound might be brought down in volume and the quiet one NOT affected, making the loud and quiet sound equal in volume.

As far as flattening EQ goes, you'd have to have a certain piano note/frequency boosted like crazy followed by an unboosted note several notes away for this to occur, the boosted note would hit, get compressed, then the compressor would stop in time for the unboosted note to pass through, effectively FLATTENING your EQ job!

But like the above poster mentioned, who the hell would do that? haha
Old 14th August 2008
  #28
Gear Maniac
 
skatingbasser's Avatar
 

I cut, compress, and boost in that order.

Why squash together some junk I don't want in there only to try to scoop it out after? Seems silly.
Old 14th August 2008
  #29
Because in some cases, having the unEQ'd signal hit the compressor will create a more interesting compression effect. There's no way to really know which way will sound better, just try them both and decide which one sounds best.
Old 14th August 2008
  #30
Lives for gear
 
synthoid's Avatar
 

Quote:
Originally Posted by tINY View Post


Actually, no. The compression part has to do with lowering gain when the input gets too hot. Without some sort of make-up gain, the tracks getting compressed only get less loud.



-tINY

Hmm? We agree that the compressor applies gain reduction when the input is too hot -- above the threshold. The question is, what does the makeup gain do? I am under the impression that in most compressors, it is simply an amplification stage at the output of the compressor -- provided as a handy way of restoring overall power levels in light of the gain reduction applied by the compressor.

If that's so, then the makeup gain doesn't affect the signal in any way except how loud it is.

But maybe you are saying that in some compressors the makeup gain is integrated in a different way into the circuit?

-synthoid
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
natpub / Low End Theory
148
brianroth / Geekslutz forum
9
dbbubba / So much gear, so little time
6
djui5 / Rap + Hip Hop engineering and production
0

Forum Jump
Forum Jump