Best compressor?

[kylen]

How about automating the volume on the offending track?

Well that's a good idea too and sometimes an option. In the case of certain transients, pulling the volume down can sound just a brutal as hitting it too hard with a limiter if the volume envelope shape isn't just right - gain reduction is sometimes too obvious in both cases.

One thing I would try then, with the volume envelope, is to make a slightly more gradual slope ahead of the transient (we're talking some double-digit milliseconds here I think) and trying to figure out a reasonable volume envelope decay (release) so the ear doesn't notice. The same thing (I think) can be accomplished by a compressor with a Knee by getting some look-ahead 'breathing room' and automating the Knee so it gradually increases and by the time the transient begins to poke its head out above the detectors' radar we're already applying compression. Then the automation dials the knee back down to the normal setting. The threshold could be automated too I think but I believe that would be more noticable.

Just tossing out some ideas here - I haven't had much experience with volume envelopes for transients but with compressorss & limiters I can usually get about 6 or 7 dB of gain reduction before my ear starts to hear the artifacts. I'll have to see how big of a transient I can push down with a volume envelope. Sometimes they can be over 12dB - which is a recording mistake in most cases.

For this kind of transient 'repair' or smoothing I use a compressor in 'limit' mode so there's is no increase in gain of the lower level signals - just limiting at a lower compressor ratio like 5:1 or 10:1. Voxengo Polysquasher and Kjaerhus GPP1 (Golden Peak Pressor) are the usual candidates although someone mentioned one here today - Euphonic and the GVST comp that I haven't tried yet.

Good idea about automation of volume envelopes, and compressor control too in certain cases I think!

How large of a volume spike do you usually push down using envelopes?
Thanks!

[bmanic]

Regarding compressors, I'll be receiving my 'holy grail' compressor soon (and EQ). The Sintefex fx2000. So I'm calling all developers, I've got one box soon that does sound absolutely fantastic and can do a lot of different sounding compression. I'm still a plugin junkie and will always be so I hope some developers will take this opportunity coz I'll be analyzing the shit out of this thing until I get plugins that sound equally good.

Right now IMHO the best native plugin compressors are (in no particular order, they all complement each other):

Kjaerhus Golden Compressor
Sonalksis Compressor
Voxengo Crunchessor

Then there are the "can be used for radical shaping" ones like Neodynium and the soon to be released Compadre (love this one!).

Am I a compressor nut? Well.. as I'm spending nearly 3000 euros on a hardware box mainly for compression, then.. u bet!

Cheers!
bManic the soon to be compressoManic!

[Kraku]

Am I a compressor nut? Well.. as I'm spending nearly 3000 euros on a hardware box mainly for compression, then.. u bet!

Holy shit! I take it you ment that literally?!?!?

[RandyHancock]

How large of a volume spike do you usually push down using envelopes?
Thanks!

I thought about it more, and first off, isn't this what a de-esser is for? You should be able to tune in the problem frequency and have it compress just those spikes and not effect the rest of the track.

Automating a compressor just seems conterintuitive to me since all it really is is an automatic volume controller. Although I do agree that it could work.

Other than that, I think what I would actually end up doing is opening a destructive editor, hilighting just the offending wave, and decreasing the volume until it sounded right. As long as you leave "snap to 0" on you should be able to reduce just one wave or two without any artifacts. When I'm saying "wave" I mean one cycle of a signal. I used to do this in cool edit if I had any pops in my dj mixes. It seemed to work pretty well, although it's time consuming. I think you'd get better results than trying to do an envelope curve in a sequencer. It would probably be hard to get the curve right like you said.

[[ Lex ]]

Regarding compressors, I'll be receiving my 'holy grail' compressor soon (and EQ). The Sintefex fx2000.

Ohh yes, those sintefex got something holy going... thats a "hell" of a gear to own.

Im just curious, did you demo/hear the Liquid Channel and so why did you go with the sintefex ?

/ Lex

[citizenchunk]

... however, it is the sidechain circuit/algorithm that determines 99% of a compressor's sound, and that is what separates the good, the bad and the ugly.

chunk - I don't write compressor algorithms - could you say a few more words about what you mean there and is it about compressors in general or about Vanilla ? Thanks!

sure.

as you might already know, every compressor is comprised of a sidechain and a gain reduction stage. at the input, the signal is split off to the sidechain path. in the sidechain, the signal analyzed for level (RMS averaging or peak rectification) and sent to an "envelope detector". the envelope detector basically just slows down the level transitions, so as to smooth the envelope. it eases into attack, and eases out of release. the output of the detector is then fed to the gain reduction stage (which in a "feed-forward" analog comp would typically be a linear VCA), which applies gain reduction to the input, based on the envelope.

(what i've just described is a "feed-forward VCA compressor," which is different from a "feed-backward opto (or valve) compressor".)

the reason for the envelope detector is simple: sudden gain changes result in intermodulation distortion. to experience this first-hand, take a non-look-ahead compressor (i.e., not a limiter or loudness maximizer) and set it for moderate to heavy compression (4-8dB), with attack/release at 10ms/100ms. listen to that. now turn the attack/release all the way down, as low as they go. (1ms is low enough.) hear the distortion? also listen to the bottom end. notice how it dissappears when the release goes down? that's IM distortion, more or less.

now, as i said before, the sidechain is 99% of a compressor's sound, and what separates good compressors from bad ones. why?

well, with a feed-forward VCA compressor (which is pretty much the standard for the modern compressor), the gain reduction transform is linear and, hence, predictable.

(NOTE: feed-back opto compressors are non-linear by design and are, hence, unpredictable. this is why they traditionally don't have threshold/ratio controls but rather "compression" controls, which basically just control the level of the sidechain signal.)

(NOTE: feed-forward "soft-knee" plugin compressors also have a non-linear compression curve, but are usually linear in time--meaning, the current compression is dependent on the current sample value, not previous values. therefore, they are predictable.)

but, if the gain reduction is predictable, then what is the unpredictable element that makes comp X sound different from comp Y? the sidechain. the sidechain determines how the compressor changes gain reduction which, in essense, is all the compressor does.

if you want to get real technical, the sidechain behavior actually influences the type and amount of harmonic distortion that is introduced by the gain reduction. that's why every compressor sounds different, and why so many compressor plugins don't sound as smooth and sweet as ChunkWare Vanilla Compressor. [ahem]

== chunk

[RandyHancock]

Oh, and the best compressor IMO is the Sonalksis one. The GCO and Crunchessor are very good, but the Sonalksis is a step up. I recently demoed them all because of the deal they had going and decided to hold out for the Sonalksis. GCO was able to match the Sonalksis at certain settings, but when I played around with things, the Sonalksis had a more consistent frequency response and held the detail of the source better than GCO. GCO is amazing for the money, but it still doesn't quite match up to the Sonalksis IMO. Crunchessor definitely felt like more of a channel comp (duh, that's what it was programmed for) but I rarely need channel comps. If I threw complex material at it, it kinda fell apart, at least compared to GCO and Sonalksis. I might still pick it up though at some point considering I might want that sound and it's only $50 anyway. I like the selectable coloring it does.

I should also mention that waves rcomp is still quite good. I included it in some of my comparisons and I'd put it ahead of crunchessor for complex material, but behind the other two.

I haven't tried the DSP stuff. That might be the best but I wouldn't know.

[jens]

Most compressors do the same job, yes. But HOW they do it - and how they sound - is very different from compressor to compressor. Sure, making a basic compressor is not that big a deal...it's making it sound good thats hard. There's an almost endless amount of small tweaks when developing a compressor that determines how it performs; envelopes, timing nuances, and the whole sidechain/detector circuit...to mention a few.

er... - so why don't you release one on your own, if you know that much about it?

[kylen]

I thought about it more, and first off, isn't this what a de-esser is for? You should be able to tune in the problem frequency and have it compress just those spikes and not effect the rest of the track.

I've used db-audioware de-esser in cases where the resonance or transient is really narrow. Sometimes they're not so then I pull out Soniformer a 32band compressor. I can get about 5 - 7 dB transparent gain reduction that way for wider bands with Soniformer, more with the de-essor on narrower bands.

Automating a compressor just seems conterintuitive to me since all it really is is an automatic volume controller. Although I do agree that it could work.

I've just started doing this now that Blue Cat has released a peak envelope recording tool called Digital Peak Meter. I can do stuff like automate the knee ahead of time or set any knob to look-ahead or ride any envelope or inverted-envelope I like. That gets me the ability to better handle the > 7dB of limiting I need to repair stuff sometimes.

Other than that, I think what I would actually end up doing is opening a destructive editor, hilighting just the offending wave, and decreasing the volume until it sounded right. As long as you leave "snap to 0" on you should be able to reduce just one wave or two without any artifacts. When I'm saying "wave" I mean one cycle of a signal. I used to do this in cool edit if I had any pops in my dj mixes. It seemed to work pretty well, although it's time consuming. I think you'd get better results than trying to do an envelope curve in a sequencer. It would probably be hard to get the curve right like you said.

Yes I could use the volume envelope on pops - impluses in other words - my transients are longer in the case I'm talking about maybe about 100-200ms but they ramp up real fast go up 12dB above the average envelope then settle back down. That's why I was trying to get a compressor in there to help me so I don't have to sweat envelope rise times so much.

Not trying to take your thread too far OT Cyprus - just trying to introduce some automation and any potential uses - and learn some stuff.

Thanks for the compressor lesson chunk! I'm reading & learning now!

[RandyHancock]

kylen,

Sounds like you've got a pretty good method of dealing with your issue. It seems like automating the compressor like that would cut down on the fiddling time of manually doing it with envelopes. I can't think of a much better solution for quickly dealing with the problem you described. I'll definitely try it if I end up with a similar problem.

Ok, I'll stop this thread hijacking now.

-Randy

[bmanic]

Am I a compressor nut? Well.. as I'm spending nearly 3000 euros on a hardware box mainly for compression, then.. u bet!

Holy shit! I take it you ment that literally?!?!?

Yep.. will have it sometime in late february if everything goes according to plan. As we live in the same city you're more than wellcome to come and listen to it.

- bManic

[bmanic]

Regarding compressors, I'll be receiving my 'holy grail' compressor soon (and EQ). The Sintefex fx2000.

Ohh yes, those sintefex got something holy going... thats a "hell" of a gear to own.

Im just curious, did you demo/hear the Liquid Channel and so why did you go with the sintefex ?

/ Lex

I've spent about 5 hours with the liquid channel and while it's a fantastic unit it's only MONO and mainly a pre-amp. I'm not a mic pre fanatic at all. The sintefex fx2000 I've got about 12 hours of experience with and it's the best sounding piece of hardware that I've demoed yet (well, the Millenia TLC-2 opto fet/tube compressor was nice too but as it's sampled by the sintefex I'm all set ).

I'll start a new thread with tons of audio examples once I have the unit. I will probably be getting a demo unit (prolly a FX8000) this month and will be sampling as much equipment as possible during that time.

I just bought a new soundcard to operate with the FX2000 smoothly, a RME HDSP 9632 with the AES/EBU balanced option.

Cheers!
bManic

[Rune L-H]

er... - so why don't you release one on your own, if you know that much about it?

Good point I'll get back to working on it then. Just wanted to point out that developing a compressor is not trivial at all....especially the part where you schedule the release date is kinda hard

[kylen]

kylen,

Sounds like you've got a pretty good method of dealing with your issue.

Thanks for thinking about this and the chat - the wife starts getting a little glossy-eyed when I start yackin like that...

Looks like the Sonalksis smokes - it's just above my price range...too bad for now

[Prefab doubt]

My choice:
Uad 1
Sonalksis