Post
by mcaff » Sat Nov 03, 2007 8:50 am
JJ, I love the idea that a measure of credibility is the gear one owns. Does that mean a gut who works exclusively ITB can never have a valid opinion on compression or do plugins count - I have 1176 as plugins. If this is not a joke and it's really important to you, I can tell you that I did have a pair of the 1176 reissues for quite a while. I loved the way the compressed, but I never really liked the tone of that particular pair, so I sold them and replaced them with a second pair of Distressors - which I agree, are not 1176s and that's why I have them instead of the 1176s. At some point, I expect I'll end up with a pair of the blue stripe 1176s, which while I've never used, sound like they're more what I'd hope to get from an 1176.
As far ash your other comments, you're way, way off the mark. But if you've got your mind made up and this is some kind of personal thing, which it sounds to me like it is, don't bother reading the rest of this post because it's just going to piss you off more.
You are not the first person to miss-read the sentence that used the term hack and in an industry that is essentially based around computers, it never occurred to me that engineers would ascribe the wrong meaning to the word - there are several, the meaning your ascribing to it, "a cab driver", and it is commonly used by computer programmers to describe temporary fixes and jury rigged workarounds when fixing software problems and that's the way it's used in the article because it' is a far more accurate and effective term than "band aid" or "short cut".
From Urban Dictionary.com:
3. To jury-rig or improvise something inelegant but effective, usually as a temporary solution to a problem
2. A temporary, jury-rigged solution, especially in the fields of computer programming and engineering: the technical equivalent of chewing gum and duct tape. Compare to kludge.
n2. This subroutine is just a hack; I'm going to go back and put some real code in later.
The sentence in the article read "I usually consider parallel compression a "hack".
That does not mean that I consider the people to use it to be "hacks" and that it's nothing more than my opinion as well as the fact that it clearly goes on to explain the specifics that that opinion is in response to. As far as the skills of the engineers that I hang out with, I don't hang out with many, and they have a wide range of skills, many are just as you describe them.
The issue I have with parallel compression, is that it allows these particular individuals to be dishonest with themselves because they mask their bad compression with the dry signal. They get all the subliminal aspects of bad compression - the agitation from the intensity of every little sound being brought up in volume, but they don't hear the transients being removed. Another way to look at it is that they're able to use peak limiting when compression would be more appropriate and the most obvious cue that the peak limiting is not providing what they want is the squashed transients. I've got a perfect example, but it's not a story to post publicly.
What I can't believe no one has pointed out yet is the hypocrisy of the comment. I thought people would find that genuinely funny to go from saying that parallel compression is *a* hack, to basing an entire technique around it. Do you see the irony?
As far as your concerns about multi-buss compression, the idea in this article has precisely zero to do with multi buss compression. The fact that you think that, indicates that you don't understand one or both of the techniques. Using two pairs of busses to mult a signal is not Brauer's multi-buss technique, so I've been racking my brain trying to figure out why you might think that and the only thing that I can come up with is the use of letters to reference stereo buss pairs. Brauer uses letters because those are the names of the busses he uses on the SSL 9k. I use letter because most consoles don't have both tracking and specialized mix busses, they just have the tracing busses that can be used during mixing. Those are usually labeled as 1/2 and 3/4 and we're I to refer to "buss 1" when discussing a stereo pair, that could be misinterpreted as the mono buss of buss 1/2. If I referred to "stereo buss 1" that might seem to imply that the console had more than one master stereo buss, so letters are the best choice and the use of letters as a generic variable does not make it plagiarizing.
I don't think I've ever used the word "expert" to describe my knowledge of Brauer's technique. I have had the unique opportunity to spend many, many hours talking to him about it and sorting through a 50,000 word transcription of a 10 hour conversation. That certainly gives me a bit of familiarity. Brauer must think so because he asked me to help him with his last Tape Op demo because I've spent more time playing with the idea on a non-SSL console than he has. Based on his comments after that demo, I'd say that I can adequately give you a general overview of the technique and explain the problem that it solves.
The concept with "Multi Buss Compression" in the definition of Brauer's mixing technique is to create multiple compression subgroups and compress within the subgroups so that the different parts of the mix don't create ducking effects when they pass through a stereo compressor on the stereo mix. An example of the problem this solves is that it's not uncommon to be nearly complete with a mix and the decide that you need to turn the bass up. Now, that level change causes the vocals to be ducked in some spots where it shouldn't be. There is more than one way to solve the problem, and multi-buss is one of them. When you group bass and drums separate, from guitars, separate from vocals and the compress them all separately, if you push the bass up, it's not going to to hit the compressor that the vocals are going through and it won't be able to duck them. That's a simplified overview of the technique. He has many creative techniques that he does within the groups, and, also uses group assignment as part of the mix esthetics.
The idea that I tried to communicate (and judging by the number of emails I've received thanking me for the article, I'd say it was more successful than not) was a method of narrowing dynamic range without hearing compression artifacts. I described it in two parts, because while it's the exact same thing being done, I still see it working in two separate ways. I'm not sure if you saw the video which used multiple busses for the routing, but that's a great example of how this is not multi-buss compression in Brauer's sense. There were multiple busses used, but they were to create a single subgroup. The alternative would be to use a mult and potentially have loading issues depending on the gear and how you do it.
The premise is that "the hack" really only has the effect of raising the low volume signals because the dry signal is unrestrained by gain reduction - this is why parallel compression is sometimes referred to as "upward compression". So if you have parallel compression in front or regular compression, you'll only hear the parallel compression alone when the dynamics are so low that the uncompressed signal is masked by the compressed signal. Assuming that you don't have you second compressor in the routing set with a threshold that it's kicking in at this point two, the parallel compressor is now taking care of the compression during the low dynamics. When the dynamic changed, it's masked and "useless" but the the sum of the compressed and uncompressed signal, which is really basically the uncompressed signal, hit the second compressor and that can be set that it's only affecting the loud sections of the song. So it's like having one compressor for the quiet sections and one for the loud sections. This is totally uninteresting unless you've had the experience of setting a compressor for a quiet verse and that perfect setting trashes the chorus - the perfect setting becomes "bad compression". If you don't hear "bad compression" or consider big verses and small choruses to be a problem, than this technique is a waste.
People have a hard time learning to hear compression. There are a few people who've said that the sidebar on "bad compression" alone made the piece valuable to them. As msmith said, everyone is works different and it's possible that all this is basic stuff to you and that relative to your skill level, this is a beginner's article.
The other concept, is to use "the hack" to bring up the low level signals so that you're sending a much narrower dynamic range to your "main" compressor. Where are all the comments in response to the idea that a compressor is best off seeing a dynamic range of 3db or smaller at it's input? What kind of asshole writes something that absurd? 6db is pushing it and beyond 10db of dynamic range going to an compressor input sounds bad? He's nuts, compressors are meant to narrow dynamic range not have it narrowed for them! What's he using, a diva compressor that has to have it's work done for him? Why doesn't he use some real compressors like an 1176....? (oh wait, we did get that comment).
I tried to get opinions from a few designers to include in the article. One understood the question but said he had no idea. Dave Derr pointed out that you can get a great sound out of a Distressor with 30db of gain reduction happening. But, you can get 30db of gain reduction on a signal that has a far narrower dynamic range than 30db. So that's not the same point. I didn't reach Greg Gualtieri until after the deadline for the article. He agreed with the numbers. He also explained that classical engineers used to use nearly this exact same technique 40-50 years ago.
So, a vocal is a great example. If you set the compressor for the verse, the chorus is going to get squashed. One option, is to have a fader infront of the compressor and ride it into the compressor - push it up of the quiet sections so that every part of the performance is hitting the compressor at as close to the same level as possible. One could argue that this is a better technique, but these days, not everyone has a fader available. So parallel compression in front of the compressor can be used if you don't have the option of a fader, and now the "hack" is being used for what it actually does - bringing up the quiet sections without affecting the loud ones.
All that's left ot explain is how to make parallel compression - as someone did very well a few posts back. You use console busses, mults and a summing box, or as another person mentioned "why not use a compressor with a mix knob?" - the answer being that there aren't many that have one. I don't know of any specific plugins that do, but I've heard there is one. I know the Chandler Germanium and the Tonelux compressors have one, but I'm not clear if the Tonelux compressor is available. I've heard there's one other hardware one, maybe a Cranesong (?) that that has a mix knob built in. The article did cover that at the end, but most time was spent explaining techniques that could be used by more people than the few who have that specific gear. I also covered how to do this with plugins for people who are exclusively ITB. I've since decided that the explanation provided earlier in this thread about assigning a group to one buss pair and creating to separate auxes fed by that same buss pair is an easier method than sending to two separate busses. That's kind of redundant unless you want certain routing options, like to have tracks feed the compression, but not feed the dry path.
I hope that this clarifies that hack was being used as a noun and not an adjective to describe the people who use parallel compression (I hope that people can see the ironic humor too). This should also make it clear that using multiple busses to send to compressors does not make something the same thing as Michael Brauer's Multi-Buss Compression - check out the Tape Op interview with him again for some real detail. I hope it's also now clear that "two stage compression" - daisy chaining a compressor with a mix knob or using busses as mults to achieve the same thing has absolutely nothing to do with Multi-Buss Compression (unless for instance you use the daisy chain set up within Brauer's Multi-Busss routing, which works phenomenally well).
