EQ, compressing - in what order?

[littledog]

Like you, Sonusman, I don't tend to use multiband compression that much because it's not a tool that I have a ton of experience with. But I know at least a couple of mastering engineers that work wonders with them, so I think their value is directly proportional to the skill of the user.

If the peaks turn out to be fast transients that spike way above the average program, it shouldn't be that hard to fix. Any compressor should be able to do the job, once you find the right threshold (high enough so that the average program level doesn't trigger it) and attack-time setting (pretty quick). Use a fairly high ratio and you should be able to create a LOT of headroom fairly transparently. The advantage of a multiband is sometimes you can get even more transparent compression than with a full-bandwidth compressor. This is one of those situations where a spectrum analysis program can be very useful in spotting the problem frequencies, which, as you pointed out, may be in the harmonic overtones as well as the fundamentals.

In a DAW system, you can even try drawing the peaks down with a pencil tool. But be real careful to work on a COPY of the file, since often that type of editing is destructive, and you may not like the results.

Personally, I've never met a "normalize" algorhythm I've really liked, so that's one step I'd avoid. There are plenty of other ways to get "louder" tracks.

 

[James Argo]

James Argo: I'm in here every day. I know these people are good and respectable, I don't know about you though, don't think I have seen you around here before...

You took it too hard... I'm there at Dragon Cave.

Maaan... ed your story has touched my heart. My tears are drop by now. I can't believe I readed those post over and over. Anyway, I used Multiband compressor sometimes. And it does a good things for it's concept. Compressing an audio signal at certain range of frequency rather than the whole signal. If you can imagine and have a better picture of what it does exactly, then you'll probably have a better though about how to use it. I know it's kind of rare to find any engineer will have it on their studio. They don't get used to it, or they very seldom find a problem that should be fixed by this device. But hey... lets open our eyes man. It's there to help us. We just should take a curve to dig more about the advantage of using it. When, Where, and How. No, I'm not a mastering guy who knows everything I'm talking about. I go Shailat most of the time I get trouble with how the compressor will work anyway...

See, I remember about the mixed audio material and a picture shot and printed out. Once you take a shoot of photograph, printed on the piece of papper, there you go. You cant erase the picture of somebody there, to expect seeing what's on a table behind him. You simply can't. Photoshop either will do. You can manipulate 'em, but not actually see what's behind him. Same thing goes to mixed tape. Once it's put there, there you go. You can master it whenever you like, but MASTERING wont take a place of MIXING. Whatever should be done in the MIXING cannot be done in MASTERING with same result. ...aaah like ed has mentioned before. It's always better to make sure you make alternative backup of your DAT. I ALWAYS make at least two version of mixing whether I can. One version of what the producer like, one version of what I like. You can also make it three or four version, it takes only tape. Tapes are cheap now (compared to your valuable data). Put the "my taste" mixed DAT for myself. Whenever the producer /musician/ band realizing something was wrong with their mixed DAT, they can always contact me to see if "my taste" one can help. Send them one copy. I can't give any solving by now. But giving my oppinion leads you to be more carefull in future I hope has a meaning for you.

 

[dobro]

"If the peaks turn out to be fast transients that spike way above the average program, it shouldn't be that hard to fix. Any compressor should be able to do the job, once you find the right threshold (high enough so that the average program level doesn't trigger it) and attack-time setting (pretty quick). Use a fairly high ratio and you should be able to create a LOT of headroom fairly transparently. The advantage of a multiband is sometimes you can get even more transparent compression than with a full-bandwidth compressor. This is one of those situations where a spectrum analysis program can be very useful in spotting the problem frequencies, which, as you pointed out, may be in the harmonic overtones as well as the fundamentals."

'once you find the right threshold'

Well, how do you do that? These days, I'm compressing whole tracks, and I'd rather use a multiband sometimes. The way I judge how much I've compressed the track (I find it *really* difficult to hear compression) is to look at the peaks of the waveform. How do *you* do it?

 

[littledog]

Same thing goes to mixed tape. Once it's put there, there you go. You can master it whenever you like, but MASTERING wont take a place of MIXING. Whatever should be done in the MIXING cannot be done in MASTERING with same result. ...

Logic says that's true, and in general perhaps it is. But then you run into a REALLY good mastering engineer and you hear the magic they can accomplish and you start to wonder...

 

[littledog]

'once you find the right threshold'

Well, how do you do that? These days, I'm compressing whole tracks, and I'd rather use a multiband sometimes. The way I judge how much I've compressed the track (I find it *really* difficult to hear compression) is to look at the peaks of the waveform. How do *you* do it?

This is where the Gain Reduction meter comes in very handy. They can be found on almost every hardware and software compressor. Often in hardware pieces there is a switch that toggles the VU meter function between output volume and gain reduction.

And then there is a strictly analog device you can use - your ears. Explanation follows:

To set a threshold in the situation I described earlier (reducing quick transient spikes that are much higher in volume than the average program material) using the gain reduction meter - watch the meter and set your threshold so that the "average" levels of the program are not triggering any activity on the meter (or maybe just a dB of reduction at most).Keep lowering it until the average program level starts to create activity on the GR meter. Then raise the threshold back up just enough so that the activity disappears. You have just set your threshold. Now only the spikes you are trying to control will set off the GR meter, assuming you don't have too slow of an attack time, and you have a greater than 1:1 ratio. Increasing the ratio will increase the amount of gain reduction on the spikes. You will want a quick release time so the compressor so that the spikes don't trigger the compressor into accidently compressing some of the "average" program that immediately follows a spike.

To do the same thing using your ears, you will need to use the bypass switch. As you lower the threshold, keep switching the bypass in and out until you hear the average program material get softer when the compression is engaged. Then slowly raise the theshold until you just get to the point where you can't hear the difference in the average program whether the compression is in or out. That's your threshold setting. Then continue as above.

GR Meters are a little easier and more accurate to use, but there's nothin' wrong with good old ears.

Any questions?

Any answers?

 

[sonusman]

Errrrrrr....ummmmmmmm...ahhhhhhh....

A couples of posts here have sort of hinted at the fact that I don't have much experience with, or possibly am not in the know of what multiband compression is/can do.

I have given multiband compression a VERY fair try. I understand it's concept quite well, and can perform some nifty tricks of my own with them if need be. Thanks

If need be...that is the question.

I didn't offer a solution to this guys problem because the only real good solution is one that he probably doesn't want to hear. REMIX THE FREAKIN' TUNE!!! Anything done in "mastering" is going to probably do more harm than good when his described problem is the case. You have a broadband problem of volume, NOT a frequency dependent problem. So, multiband compression will do little. Think about it. If the instrument covers a big range in the sonic picture, multiband compression is probably going to be less effective than broadband compression will be.

Either type of compression though is not going to change the mix. If the drum loops, a broadband instrument I will assume, are too loud compared to the other instruments, then it is just plain too loud. No amount of compression, eqing, ANYTHING is going to change how that. Sorry.

For broadband gain reduction, use broadband compression. For narrow band gain reduction, use a multiband compressor on the band that needs gain reduction. It is that simple. I think everybody get's that. But when you have a case where something is just plain too loud in the mix, and it is broadband, broadband compression will turn EVERYTHING down. So, the relative levels in the mix will be the same during gain reduction, it will just be the same mix at a lower volume (unless make up gain is used) The guys loops "peak" 10dB louder than the rest of the mix. I am sorry, I have never heard ANY limiter applied to broadband signals do 10dB of gain reduction without some SERIOUS bad artifacts to the sound (read distortion and/or pumping). Now the question is, is that a more desireable effect than just leaving things alone? I suppose if your applied the compression/limiting/eq creatively there could be some merit. But if preserving good audio is the goal and you are expecting a dynamics processor to change the relative level of instruments in a mix, well, it is just not going to happen.

Ed

 

[pipelineaudio]

Call a multiband compressor a "multiband limiter " and all of a sudden you have a VERY useful tool. Call it a "de-esser" and you got a classic

 

[dobro]

Pipeline, I tried using my multiband for de-essing, but I found it really difficult to find the offending frequency. I know how to boost and sweep with a narrow Q, but I wound up having to cut most everything between 7K and 8K, and then it affected the sound a bit negatively, the way Ed described in a previous post. How narrow a range of frequencies do you typically work with using a MB for de-essing?

dobro

 

[dobro]

littledog - thanks, that was useful - so far, I've ignored the gain reduction meter on my compressors. But yes, I've got a question, you talk about homing in on the average level of the material and then setting the threshold just above that. Okay, but how do you decide what the average level is?

 

[scottboyher]

!

I believe if the situation is as I have read it, it would pay to re-record the tracks.. Don't try to fix something that can't be fixed like I do.

 

[littledog]

littledog - thanks, that was useful - so far, I've ignored the gain reduction meter on my compressors. But yes, I've got a question, you talk about homing in on the average level of the material and then setting the threshold just above that. Okay, but how do you decide what the average level is?

Reread my post - I described how using either gain reduction meters or ears. Just lower the threshold until the GR meters start jumping, then raise it back up until they more or less stop - that tells you that you are now "above" the average level of the song. If the song has drastic differences in volume level from section to section, then you may have to set the threshold levels one section at a time. Try it. It's a lot easier to do than to think about.

P.S.to Sonusman: Sorry if I implied that we shared a common ignorance or incompetence. Regarding your last post, I would have to hear the material for myself before deciding it's as totally hopeless as you make it out to be. And Pipeline's points are well taken.

 

[pipelineaudio]

Dobro, some multibands are better suited than others for this. But remember, a De-esser IS a multiband, its just only got one band going A normal de-esser like a DBX has a pretty damn wide range, even when tuned sharp.

With a waves C4, for doing se-essing, it only really seems to work in REALLY hard limiting for me, unlike a normal de-esser which isnt so offensive if its actually compressing. The waves stuff seems best as a de-esser or de-popper if you do like littledog says and just use it to lopp off the REALLY offensive peaks

 

[James Argo]

Logic says that's true, and in general perhaps it is. But then you run into a REALLY good mastering engineer and you hear the magic they can accomplish and you start to wonder...

They still don't REMIX the song. They do other way to make the song a bit more "safe" and interesting. Yes, they do their magical hocus pocus to "fool" the listener. And it makes the song better than before. But it doesn't bring the -too loud- instrument back without affecting the rest in it's freq range. ed figured it clear. However, I agree some mastering guys makes the song alot better to be heard, even when the mixed was crap before. I'm gonna make sure to watch the spectrum analyzer carefully. Damn... I can't believe I get paid for this...

 

[dobro]

littledog - sorry, I still don't get it. If I set the gain reduction meter so that it stops jumping, then that means the compressor's doing no work at all on anything, right? So, what does that accomplish?

I tried it yesterday, the way you described. When I applied that setting, it changed the waveform only a very tiny bit, just as you'd expect. But I'm looking to *tame* peaks, not tickle 'em under the chin. LOL

 

[littledog]

littledog - sorry, I still don't get it. If I set the gain reduction meter so that it stops jumping, then that means the compressor's doing no work at all on anything, right? So, what does that accomplish?

I tried it yesterday, the way you described. When I applied that setting, it changed the waveform only a very tiny bit, just as you'd expect. But I'm looking to *tame* peaks, not tickle 'em under the chin. LOL

It seems it might almost be easier if I just flew out to Singapore and showed you. You set the THRESHOLD so that it only starts compressing when the obnoxious peaks come in. That means the average program is unaffected. But once the threshold is crossed, there is no limit to how much compression you can apply by using a very fast attack time (2ms?) and a very high ratio (20:1?). That hardly qualifies as a tickle.

But to be honest, and I'm not saying this as an insult, but only as a practical matter: we seem to be going round and round on this, and if it is going to take you hundreds of hours just to figure out how to use your compressor, maybe it would be far more cost and energy effective to hire an experienced engineer in your area to show you how.

It's really not that difficult in theory, but you seem to be having some sort of mental block. Hey, it happens to the best of us from time to time...

 

[dobro]

"It seems it might almost be easier if I just flew out to Singapore and showed you."

Sure, okay. There's a good time to had out here too. It doesn't *all* happen in the US.

"You set the THRESHOLD so that it only starts compressing when the obnoxious peaks come in. That means the average program is unaffected. But once the threshold is crossed, there is no limit to how much compression you can apply by using a very fast attack time (2ms?) and a very high ratio (20:1?). That hardly qualifies as a tickle."

Uh-huh. But before you were talking about setting the threshold to deal with the 'average' level. Average level and obnoxious peaks are not the same thing, right?

"But to be honest, and I'm not saying this as an insult, but only as a practical matter: we seem to be going round and round on this"

No, we're not going round and round. You've responded to my question, and I've asked a subsequent question. That's exploration, not going 'round and round'.

"...and if it is going to take you hundreds of hours just to figure out how to use your compressor, maybe it would be far more cost and energy effective to hire an experienced engineer in your area to show you how."

Hundreds of hours? No, I don't think it will take that long. As for getting an experienced engineer in my area to show my how, you seem to be laboring under the impression that only engineers are capable of teaching homers how to record and mix. But I think that's probably not true, considering that you haven't been able to teach me what seems to be a very simple aspect of compression....

"It's really not that difficult in theory, but you seem to be having some sort of mental block. Hey, it happens to the best of us from time to time... "

Yeah, I know. I'm a teacher, and I know all about mental blocks. I also know that there are people who know stuff but don't know how to teach it. You know, I've got a flute teacher, and he says stuff to me like: "I know you're not getting it, but I don't know how to explain it to you so that you'll get it." He's a second best sort of teacher, so I stick with him. (The best ones are the ones who can explain things so that it's understood. It's a skill and an art.) As for the ones that aren't able, or don't take the time, to explain simple stuff to me, I know what to do about that too. I say stuff like: "Hey, it happens to the best of us from time to time..."

 

[sjoko2]

Dobro, I think you'd be well advised (so would everyone, apart from the few who can hear and name each frequency perfectly), to buy a cool analysing software package, like for instance spectrafoo

 

[dobro]

Yeah, I'm learning how to use Cool Edit's frequency analyzer - it's showing me all sorts of things going on in my tracks. I'm not experienced enough to hear a track and know: "Oh, the problem's the E and the B and their overtones." But the analyzer shows me that.

Having said that, I'm not experienced enough to know what to *do* about the track with the hyped E and B and overtones. But it gives me a useful place to start tweaking and experimenting and listening to the result.

 

[sjoko2]

So, if you can see where your peaks etc. are, you will now know where to apply your multiband, right??

 

[dobro]

Yeah, of course - that's easy to understand.

What I can't understand is what littledog tried to explain to me - something about setting a compressor above the 'average' level of the track, so that it would nail the peaks and spikes above the average program. That sounds useful to me, but I can't figure out how to determine the 'average' level of the track with my compressor - it's triggered by anything that crosses the threshold I've set, so I can't find the average level using the gain reduction meter. Either littledog has some special compressor or he's talking about some arcane engineer thing that nobody else knows about. Well, maybe the Masons know about it. But probably not even the FBI.