Check out the Bruce Swedien sessions on GS..

  • Thread starter Thread starter BigRay
  • Start date Start date
I agree it can be a tool but you are now using it ( a compressor) as an outboard effect box, not as a compressor for dynamic range reduction. If you at all blend the uncompressed with the compressed signal the dynamic range now remains unaltered. You are looking for the effects of compression, how it alters the sound envelope, i.e. how it changes the attack, release, rms, the introduction of transformers, vca's etc. etc. etc. Let's just call it what it is. The same would occur with using eq which is again why we don't use aux. sends with outboard eq. Let's say you eq a signal, then blend it with the unaltered original signal, you effectively undo the eq by doing this. Same with compression. However if you blend it so that the eq'd signal is very hot relative to the unaltered (so it's not very audible) then effectively you are just eq'ing the signal. Of course if the eq colors the signal in some other way, i.e. due to signal path construction then you are using the eq as an effect box. This is analogous to using parallel compression. I agree you may like the way it sounds but you are not really using it for reduction of dynamic range again unless the compressed signal is very hot relative to the original signal so the original signal is almost inaudible. If you are blending the two equally you are really just liking the sound because of the artifacts introduced by the compressor blended with the unaltered sound, i.e. it is now essentially an effects box. Nothing wrong with this but again just call it what it is.

Another example, say you have a wild snare hit. You wish reduce dynamic range so this hit is not to loud relative to the others so you compress it. Now blend the compressed signal with the uncompressed signal almost equally. The wild hit will now be back. The tone of the snare may now be altered though and you like the new tone coming back from the compressor. This could be for a multitude of reasons: the compression element, transformers, mild distortion, action upon the sound evelope due to paramter settings, tube saturation, etc. etc. Now you really dig the sound but the loud hit doesn't bother you so you blend the two signals for what you perceive to be the perfect sound: you've just use the compressor as an effects box. I do this often, sometimes I just like the sound of a 160vu on a snare so I'll patch it in and set it for very light compression (the box is fast) just because I like the sound of the action of the compressor on the signal. I've just use the compressor as an effects box without blending, not really to tame dynamic range. By being able to blend the "wet" with the "dry" you may get a little more flexibility with parallel compression but you can generally acheive the same "effect" by altering compressor paramters without resorting to blending. Each method works but it is not really compression because dynamic range is not being altered. (well in my case it technically is but not the purpose of the procedure)


oh a note on doing this with digital compression in a DAW: if you are not delaying the original track relative to the latency introduced by the plug-in on the compressed track you may be introducing a delay which causes phasing and which may in turn be perceived as "thickness" which is probably really nothing more than compression artifacts combined with delay and blended with the original.
 
sweetnubs said:
I agree it can be a tool but you are now using it ( a compressor) as an outboard effect box, not as a compressor for dynamic range reduction.
Click to expand...

Obviously, I simply don't have the ability to explain this to you, so I'm going to go play some music. Maybe Glen can explain it. Alternatively, I recommend one more time that you consult Bob Katz's fine book, "Mastering Audio." This is a useful compression technique, it is well-known in some circles, it does work and it does affect the dynamics of the track, by holding the lower levels up while leaving the peaks intact. You can label it whatever you want. I use Bob Katz's name and call it parallel compression.

Otto
 
I'm sure it is a useful tool but please don't confuse noobs by calling it compression. By the way compression can be and is used to hold peaks in place while bringing lower levels up via make-up gain. You don't need parallel compression to do this, any old compressor patched into the insert will work for this purpose.
 
sweetnubs said:
I'm sure it is a useful tool but please don't confuse noobs by calling it compression. By the way compression can be and is used to hold peaks in place while bringing lower levels up via make-up gain. You don't need parallel compression to do this, any old compressor patched into the insert will work for this purpose.
Click to expand...

Excellent! Now we are getting to the lack of understanding that I've been trying to address. First, you are confused, as are lots of folks, at to what compression includes. Downward compression is, by far, the most commonly used method, but it isn't the only one. Compression is anything that reduces the dynamic range of the signal.

More important that clearing up definitions, however, is understanding the critical difference in the actual sound between downward compression and upward compression. Sure, you can take a signal and take 15 dB off of its dynamic range using either technique, and even match up the peaks again when you're done, but the results will sound entirely different!

What we hear when you push down the peaks is way different from what we hear if we bring only the quiet parts up. Downward compression can not simply hold peaks in place and bring lower levels up. That's what upward compression does! Downward compression holds the quiet parts in place and squashes the peaks.

Let's compare the two methods.

To keep the math simple, assume you have a signal with about 50 dB of dynamic range and a peak to RMS ratio of 15 dB. You use downward compression to knock the highest peaks down by 15 dB and pull the gain up by 15 dB. In that case, you've left the quiet parts pretty much alone but compressed the peaks. Then you just turn everything up louder. That is what is typically done in the name of "compression" and it can easily wimp out the sound. Suppose, in this case, we took our signal with a 50 dB dynamic range and ran it through a compressor with a threshold at -25 dB and a 2.5:1 ratio. We'd knock 15 dB off the dynamic range by squashing down the peaks. That'll knock down the peak to RMS level a bunch. Just turning the gain up 15 dB will make things louder and you could adjust the top peaks back to 0 dB, but it will sound very different from what we create with parallel compression. You've lost the lively dynamics because you operated on the loud part of the signal only and left the quiet parts alone, then just turned everything up loud.

Now let's see how parallel compression, an approximation of upward compression, works:

Take that same input signal with 50 dB of dynamic range run it into the compressor along with the uncompressed signal and then mix together with some gain on the compressed signal. Set the compressor threshold at -50 dB and the ratio at 2.5:1. With no makeup gain, the quietest input at -50 dB will come out of the compressor at -50 dB, just like the uncompressed signal, but the loudest peak input at 0 dB will come out of the compressor at -30 dB, because it is only allowed to rise 50 dB/2.5 = 20 dB. So the mixer output from the loud 0 dB signal will be almost exactly heard at 0dB and almost exclusively from the uncompressed signal (the -30 dB compressed signal is negligible). Now let's put, say, 15 dB of gain on the compressed signal. The compressed signal from the loudest peak 0 dB input is still only at -15 dB, so the output is still pretty much right at 0 dB and still almost entirely from the uncompressed signal. Indeed the parallel compressor pretty much leaves all the louder components, not just the top peaks but most all the louder parts, unchanged. But the quietest -50 dB input into the compressor comes out at -35 dB because of the make up gain. So the output from the mixer for the -50 db input level is about -35 dB and comes almost entirely from the compressed signal.

So, we've left the peaks pretty much as they were in the uncompressed signal, but the quietest parts are 15 db louder (at -50 dB input). The peak to RMS levels won't be affected much, so a lively, dynamic recording will still sound lively and dynamic and have that same peak to RMS ratio of 15 dB, but the dynamic range has been pinched by the same 15 dB we did with downward compression and the quiet parts are easier to hear (say in a truck on the highway!).

Is the difference getting clearer now?

Otto
 
Ofajen,
You have explained it better than anyone ever could hope to ! rest assured. Flat topping has become so ingrained in audio culture; your trying to hold back a tidal wave with a thimble !

I for one applaude your effort to be of service and share the information. :D
 
sweetnubs said:
I'm sure it is a useful tool but please don't confuse noobs by calling it compression.
Click to expand...
I'll send a memo along to all the Big Boys and Big Girls in Detroit and New York who pioneered this technique on vocals and drums, respectively, long before most people on this board were even born, and who's home cities have lent their name as specific types of compression methods. If you've listened to any Motown at all, you've heard parallel compression used on the vocals.

I've already offered my entry-level treatise on compression for "noobs", in which I purposely left out more advanced concepts and techniques like upward compression, range compression and parallel compression, so as not to hit them with everything at once. But that doesn't mean they can't be talked about here or that they shouldn't be called compression, because they are indeed forms of compression.

To say that just because parallel compression retains overall dynamic range that it is not a form of compression is a *very narrow* - and quite incomplete and incorrect - definition of compression, and is in fact ignoring the fact that compression is indeed happening. With parallel compression, major elements of the signal *are* being compressed. It is indeed a form of compression. There are other examples:

If a signal has a maximum dynamic range envelope defined by a heavy bass beat, and one attacks the midrange and treble with a multi-band compressor, one is applying major compression to majority elements of the signal, yet the overall dynamic range of the signal remains untouched. Is that not compression?

If one applys a multi-range compressor (not to be confused with multi-band) and applies downward compression on everything inputting between -6 and -12dBFS by 4:1 and concurrently expands everything between 0 and -6 by the same ratio, there is a load of compression going on. Yet there is no change in the overall dynamic range of the signal.

And if one is going to complain that parallel compression is just an audible effect and not "real" compression for the sake of actual range limiting, I'd reply that when one has to use compression on a bad teenage punk band (as you referred to earlier in the thread), that the use of compression to make their amateur drummer sound as if they actually know how to play drums, or to crush their guitars into a wall of sustained distortion, that those are also nothing more than audible effect uses for compression. Yet nobody seems to have a problem accepting those as legit forms of compression even if they (like I) don't particularly like them.

G.
 
SouthSIDE Glen said:
I've already offered my entry-level treatise on compression for "noobs", in which I purposely left out more advanced concepts and techniques like upward compression, range compression and parallel compression, so as not to hit them with everything at once. But that doesn't mean they can't be talked about here or that they shouldn't be called compression, because they are indeed forms of compression.
Click to expand...

Glen: Great post, as usual. Thanks for clarifying. Your examples show that even my quick and dirty attempt at a definition was incomplete. Perhaps a simple working definition is that compression is going on when you reduce the dynamic range of at least some part of a signal?

The sad truth is that I learned a good bit by forcing myself to think this stuff through in order to try to explain it, but that is often the way it works for me... I don't really understand stuff until I have to teach it to others, if even then!

Thanks!

Otto
 
flatfinger said:
Ofajen,
You have explained it better than anyone ever could hope to ! rest assured. Flat topping has become so ingrained in audio culture; your trying to hold back a tidal wave with a thimble !

I for one applaude your effort to be of service and share the information. :D
Click to expand...

(drowning sounds) gack! uhhh... thanks! going under again!... gack!... dang! I dropped that blasted thimble again... whoa, that wave is really coming fa...

Otto
 
ofajen said:
The sad truth is that I learned a good bit by forcing myself to think this stuff through in order to try to explain it, but that is often the way it works for me... I don't really understand stuff until I have to teach it to others, if even then!
Click to expand...
That's half the reason I am on this board myself. The stuff I do already know, the questions on this board keep bringing to the forefront of my mind.

When just sitting behind the circuits working under a schedule, it's very easy to fall into a standard sort of rut of standard techniques and ideas. There is so much information revolving around this business that it can be easy, if not to actually forget a lot of it, to at least let a lot of it collect dust in a corner of the mind. Fielding or reading all these questions and answers, and as you say, forcing myself to think through the answers yet again keeps all this information fresh and within arm's reach; this in turn helps keep my sessions fresh and brings tools to them that I might otherwise have overlooked.

And for the questions that I do not know or details that I'm missing, I'll research the question by old-fashoned study and/or check/test it myself on my own system. There are a few new ideas and kinks I have discovered this way.

And of course, for those ideas where there is more than one way to skin a cat or to which my experiences have not yet taken me, the opinions of many others here are invaluable.

G.
 
As far as a "definition" of compression, in this context I guess I'd think of it more as "compression techniques" than just simple "compression". Any technique that involves using some form of compression to modify the sound is a compression technique. Does the entire signal envelope have to be compressed overall for it to be considered "compression"? I think we're just playing with semantics there, but I'd personally say that is just one breed or flavor of audio compression. Probably the most common one, sure, but there are other techniques for compressing just part of the signal or to combine compression with other dynamic or signal processing to be able to compress the overall signal in a way whereas the amplitude of the overall envelope does not change.

In fact, I just thought of another one as I'm typing this. Ecktronic asked earlier for an example of expansion use. I'll go one better, I'll give a combination of compression and expansion where the desired signal is compressed and yet the dynamic range of the whole signal is not reduced: noise reduction.

This technique is one I personally stole from the idea behind old-school dbx tape noise reduction (though I'm sure many others came about this technique in other ways.) The way the original dbx noise reduction for tape worked was to apply fairly strong compression to the entire signal before recording it to tape at a level well above the level of tape hiss and system noise. Then on playback it would re-expand the played-back signal. The signal would be back to it's normal dynamics, but the tape hiss that was added after the compression would be knocked down to almost completly inaudible levels.

A slightly different version of this can also work as an excellent form of noise reduction in the studio. One or another version of this I in fact often prefer over the use of noise gating. Sometimes it's possible to really reduce noise just by running downward expansion at a very low threshold. Sometimes (it all depends upon the content) a judicious threshold can be found where what is above the threshold can be compressed and gain pumped and what's below it is expanded. What you wind up with is a signal tightened via compression and a noise floor dropped via expansion, and yet the overall dynamic range from floor to peak is the same or possibly even greater than it was before the signal was compressed.

There are several variations on this technique. What you do with the signal above the expansion threshold depends upon what it needs. Sometimes it's downward compression and gain, sometimes it's upward compression, other times it may be expansion and limiting, etc.

G.
 
SouthSIDE Glen said:
As far as a "definition" of compression, in this context I guess I'd think of it more as "compression techniques" than just simple "compression". Any technique that involves using some form of compression to modify the sound is a compression technique. Does the entire signal envelope have to be compressed overall for it to be considered "compression"? I think we're just playing with semantics there,
G.
Click to expand...

Yup .or was it NOMENCLATURE?? you say tomato I(and Dan Quail) say tomatoe! :p

How about we call it low level enhancement: oh probably to vauge or already taken!

Confused, try this.

http://www.rane.com/pdf/bewilder.pdf




Ofajen, did someone throw you a life perserver?? Ofajen,,,,,,,,,,,,, Ofajen??? oh shucks, the tide got him!!!!!!!!! :p
 
Well, the general umbrella is referred to as "dynamics processing". This includes all forms of compression and expansion, anything having to do with directly controlling dynamic range.

But yeah, you're right, flat. It's all goofball wordplay.

A Firkin of ale and a firkin of water are different, both in London and in the countryside.*

G.

* A very obscure reference that most won't get. That's OK, no problem. Nothing to see here. Move along, move along.

But those who do catch it should appreciate it as a golden nugget right on target ;).
 
Thanks for the reference. That is an interesting discussion with Bruce Swedien. But he did not say he does not use compression at all. He just does not use that much compression. I found it interesting that MJ used an SM7 for Thriller. Makes me want o buy one.
 
james hetfield also did the vox for master of puppets/and justice for all through an SM7...that REALLY makes me want one!
 
ah yeah...but I am sure MJ and Hetfiled had a hell of a vocal chain..it wasnt JUST the sm7
 
yea, i don't remember what preamp/compressor was used for those sessions - flemming rasmussen scanned and posted all the session notes on GS a while back, but i forgot many of the details
 
MJ used a Neve preamp. I have a Vintech Neve clone so that is why I'd really like to check out the SM7 with it.
 
Anyone know what kind of set up Simon Lebon used on Duran Duran's Rio record? I always liked the production of that record and the vocals always sounded great. It always sounded like there was a light flange on Simon's vocals.
 
I agree on this one. It's just a matter of semantics. Overall dynamic range is not reduced using parallel compression. However it is reduced on the channel in which the compressor is applied so compression is occuring, so I'd just file it under compression techniques. I guess I'm just a traditionalist. I generally like quick, fast, light and transparent compression when needed to reduce dynamic range. I will dabble in "compression techniques" when I feel the music i'm working on requires it. Also I do know and could care less if some "big guns" use this technique. I know plenty of "big guns" who are shitty engineers in my opinion and I have used "big gun" mastering engineers in the past who cost lots of money and some of them (one in particular) did mediocre but not awful jobs. I do enjoy motown engineering though and hold it in good regard. I have used parallel compression on rare occasion so again it's not lost on me.
 
Last edited:
You must log in or register to reply here.
Back
Top