outboard mastering levels

[FALKEN]

okay i've tried some more stuff and I am going to give it another pass, with my dbx 166XL peakstop limiter, and I am going to bring the RMS from about -20 to about -14. it should sound very nice for the style of music, where the vocals are very percussive, from what I heard toying with it.

now my question for the pros is, since I am going to try to do this outboard, what should my incoming levels be (back into the PC)? as hot as possible? peaking around -6, -12?

also how would you go about getting the levels of each song identical? digitally you can go back and forth and make adjustments...but now outboard I could try to record each song near the same volume and then normalize the whole recording at once, or perhaps each song separately? there would be no need to do any additional digital processing other than normalization and maybe dither (running at 24/44.1). any help would be much appreciated.

 

[Massive Master]

If this is the "final step" then you want your levels to be as close to "done" as possible.

If the peakstop is actually catching the peaks, bring the level up to where you want it to be (somewhere around -1.0 to -0.5 is where I'm usually at) and go. Normalizing shouldn't be an issue (don't be afraid of leaving that half dB of headroom).

Of course, your A-to-D will have something to say about this also...

As far as matching volume, you need to get a handle on what (XX-dB) sounds like when your monitor controller is at (XX) and just go. If you find a tune is a whisker too loud or too soft, just run it again if you need to.

 

[FALKEN]

good advice. that is going to make 100% of a difference.

one more question - when I adjust the volume of the songs, I should basically leave the limiter alone and adjust the volume going into it, correct? that way the peaks of each song are same volume.... otherwise I could adjust the output from the limiter..i.e., volume changes should be pre-limiting, correct?

 

[FALKEN]

and another thing...

why get the tracks as hot as possible when mastering but peak much lower when tracking? Is it just a matter of doing as little digital adjustment as possible?

 

[Massive Master]

The simple answer is keeping headroom... You always want it until the very last chance.

Personally, I'd like some AFTER mastering, but that'll never happen again I think...

There are a lot of chances to do damage - Plugs, D-to-A reconstructive distortion, outboard... If the levels are kept "cool" during all of that, it can make a fairly dramatic difference in the end result.

 

[FALKEN]

question

question about outboard levels and soundcards - I know this has been answered a million times but I don't think ever quite in this way.

-12 digital is 0 analog right? so when I send my mix out at +4 from the soundcard, unless I lower the output 12 db I am going to overload the compressor. then, after the compressor, I will need to again boost the signal before it gets recorded back into the PC, by 12 db, right?

OR, I could run my soundcard at -10 and my compressor at +4. then -10 coming out of the PC would approximate +4 at -12dB (analog level). and when the signal comes back from the compressor at +4 analog level, if the PC is running at -10, it will peak near 0 digitally.

basically, running my compressor at +4 and my PC at -10 theoretically negates two volume changes (or comes pretty close).

which should I do?

 

[TexRoadkill]

Your confusing meter levels (db) with operating levels (dbu or dbv). Keep your gear set so that the -10/+4 is consistent between the gear. Mismatching will cause distortion and/or degredation in your signal to noise ratio.

When it comes to mastering I would recomend using a computer limiter in the DAW rather then a cheap outboard unit. Limiting is one of the things that computers do well because of their look ahead capabilities.

 

[Massive Master]

True, but I would still suggest an analog limiter because of the sound and performance...

[EDIT] Okay, I saw the word "cheap" there. A *good* analog limiter. [/EDIT]

 

[FALKEN]

Your confusing meter levels (db) with operating levels (dbu or dbv). Keep your gear set so that the -10/+4 is consistent between the gear. Mismatching will cause distortion and/or degredation in your signal to noise ratio.

Yes I realize I am confusing the two ideas, I am doing it on purpose. the point is that the "actual" difference between -10 and +4 is 12 db, the same difference between "analog" and "digital" operating levels. so I was wondering if I can use that to my advantage. You are saying no but I already did it and didn't notice any distortion or degradation. my path was PC(+4) -> dbx (+4) -> PC(-10). this seemed to give the best levels throughout the chain, amazingly.

When it comes to mastering I would recomend using a computer limiter in the DAW rather then a cheap outboard unit. Limiting is one of the things that computers do well because of their look ahead capabilities.

well thanks for your recc. but I was trying something different, and I think I prefer the results so far. I haven't given it the car test but I have a feeling my 166xl kicked the crap out of my plugins.

I ran an aggressive comp with light limiting, a smooth comp with light limiting, and a balls-out limited run. I am going to give all three a listen on different systems to see which sounds best, and I might even pick different comps for different songs. Is this what MEs typically do?

 

[Blue Bear Sound]

-12 digital is 0 analog right?

Not necessarily - it depends on the gear in question... manufacturer's calibration point for 0VU can vary somewhere between -20dBFS to -10dBFS.....

 

[FALKEN]

Not necessarily - it depends on the gear in question... manufacturer's calibration point for 0VU can vary somewhere between -20dBFS to -10dBFS.....

well I've been suspecting my dbx of being calibrated funky.. the optimal levels I found might seem to confirm that. I eventually had to stop thinking about where the knobs and switches *should* be and just put them where they worked the best.

secondly, I learned an invalueable lesson that relates to why you guys are always complaining about getting levels higher, but I guess I had to figure out my own take on it. but what it comes down to is that getting the levels loud and getting the track to sound good are two totally different things.

 

[Farview]

question about outboard levels and soundcards - I know this has been answered a million times but I don't think ever quite in this way.

-12 digital is 0 analog right? so when I send my mix out at +4 from the soundcard, unless I lower the output 12 db I am going to overload the compressor. then, after the compressor, I will need to again boost the signal before it gets recorded back into the PC, by 12 db, right?

OR, I could run my soundcard at -10 and my compressor at +4. then -10 coming out of the PC would approximate +4 at -12dB (analog level). and when the signal comes back from the compressor at +4 analog level, if the PC is running at -10, it will peak near 0 digitally.

basically, running my compressor at +4 and my PC at -10 theoretically negates two volume changes (or comes pretty close).

which should I do?

I have come a little late to this.
I could be wrong, but it seems like you are mistaking peak levels for volume levels. You seem to be trying to trick your equipment into giving you 12db of gain without just turning up the volume by 12db.

The levels that you are supposed to have at -12dbfs are the average levels. Average is a really bad word to use, as is rms, because different instruments have different peak-to-average ratios. So with drums, for example, you will never get the average up to -12dbfs before you overload.

I'm hoping that you are old enough to have had a cassette deck. Remember how you were supposed to set the levels on it? You were supposed to get the needle to sit right around 0dbvu. If yours had leds, you would light up all the leds up to 0 and just let the red ones above 0 blink every once in a while. Those meters were really slow compared to what we have now on our digital equipment. There was no real way to tell where your peak levels were, so you had to concentrate on getting the needle to bounce around 0. That is what should be happening at -12dbfs.

Now, when you go out of your computer into your compressor, you should set everything to the same operating level (-10 or +4). Send your signal out to the compressor and use the makeup gain on the compressor to get the signal to average around -12dbfs in the computer. Then normalize to get the volume boost from the compression.

Why normalize instead of just boosting the compressor, you ask? (I'm psychic in my spare time) Well because, if you do not have a pristine signal path with the best that money can buy, you will be doing more harm to your signal by running your analog path 12db hotter than designed. All of your equipment was designed to run most effectively at line level. Running it that much hotter brings you closer to the limits of what it can do.

 

[FALKEN]

ok..

both of you guys are making sense but you are not saying the same thing. one guy says to get the levels "as close to done" as possible and not to normalize or "worry about the last .5 db", the other says to average where the eqipment performs at its best and then to normalize.

its strange that you say that I would have a 12 db boost, because the mix only really gained maybe 2 db or so in RMS (shocking!), and I lowered the output gain of the dbx maybe 2-3 db and the faders maybe 1 db. so 6 db must have dissapeared (cabling between mixer and gear?). also this sounded better than boosting the dbx's output 6-12 db. maybe i messed up tho and i'll have to do it again (maybe if i paid for mastering i wouldn't be banging my head on the wall every day for a month). another car test tomorrow.

 

[Farview]

ok..
its strange that you say that I would have a 12 db boost, because the mix only really gained maybe 2 db or so in RMS (shocking!),

I was using your numbers from the previous post.

-12 digital is 0 analog right? so when I send my mix out at +4 from the soundcard, unless I lower the output 12 db I am going to overload the compressor. then, after the compressor, I will need to again boost the signal before it gets recorded back into the PC, by 12 db, right?

Somehow I got the impression that you were using a DBX 166, this isn't the cream of the crop. It isn't bad for what it is, but I certainly wouldn't push it too hard. That is why I said to run everything at nominal levels and normalize. With your setup (what I know of it) it seemed like the safest way to go. With really nice equipment (like John has) there is often an advantage (or, at least, no disadvantage) to running your levels that hot. With budget minded stuff, there are a lot of pitfalls.

 

[FALKEN]

Somehow I got the impression that you were using a DBX 166, this isn't the cream of the crop. It isn't bad for what it is, but I certainly wouldn't push it too hard. That is why I said to run everything at nominal levels and normalize. With your setup (what I know of it) it seemed like the safest way to go. With really nice equipment (like John has) there is often an advantage (or, at least, no disadvantage) to running your levels that hot. With budget minded stuff, there are a lot of pitfalls.

I see your point. I am using a 166xl and although it was only a few hundred bucks, there are quite a few models "under" it, I would say its a little over half-way up the dbx food chain. it has full controls, plus automatic settings, a layered setting, a peakstop limiting section, and an expander/gate section, and very good LEDs. but I think you might be right about not being able to push it to hard. I really couldn't get any volume out of it because the compression was pretty slow and the peakstop limiting was actually very "soft". if i set it to only limit every 10-20 seconds, I got virtually no volume raise out of the track. to really clamp down on those transients, it ended up limiting other stuff, maybe every 5 seconds or so. basically I had to either over-limit or under-limit because the limiter was too soft. the compression sounded really really phat but anything more than 4 db of GR it fell apart, and even that 4db couldn't lower the overall volume of the track by more than 1 db or so. not sure how normal all of this stuff is in a pro environment.

as for the rest of my gear, I use an echo layla 24, which is my third echo interface, I have always used them and consider it very quality gear.

please let me know (pros) if any of this is normal or sends up a red flag...

 

[Farview]

DBX compressors, even the good ones, are not really the best for what you are trying to do. But if that is what you have, lets see if we can do something with it.

With improper settings of the attack and release, you could actually loose volume instead of gain it. If you have too slow of an attack, you could actually make the transient bigger in comparison to the decay. This is the opposite of what you want.

When you try to compress it harder (the only way you will be able to bring the levels up) you say it falls apart. Is that because the kick drum is making the compressor pump too much? (or something similar...any 1 instrument triggering the compressor) If it is, you might want to side chain the comp so that it doesn't react too much to the kick.

This is one of the reasons that mastering compressors are so much money. They can compress so much harder without ruining the sound. If your RMS volume of the track is -20 and you are trying to bring it up to -12, you will have to be into a compressor/limiter about 8db at the loudest peaks. That is way more than you will get out of the 166xl.

I'm not down on DBX comps, I've owned and still own a ton of them. I would never run a mix through them, they just aren't transparent enough.

 

[FALKEN]

yes I experimented with the side chain on the lows and also to squash some honking low mids and also leave the cymbals untouched. I also experimented with the "countour" button which I think acts as a sidechain on the slows. by "fall apart" I mean that the air and clarity dissapears, and the whole image becomes a blurry mess.

as far as attack time, I ran a few passes, some as quick as the comp will go and some pretty slow (like the manual suggested for a mix). still deciding which sounds better but even the fast attack didn't really get any transients.

 

[Farview]

yes I experimented with the side chain on the lows and also to squash some honking low mids and also leave the cymbals untouched. I also experimented with the "countour" button which I think acts as a sidechain on the slows. by "fall apart" I mean that the air and clarity dissapears, and the whole image becomes a blurry mess.

as far as attack time, I ran a few passes, some as quick as the comp will go and some pretty slow (like the manual suggested for a mix). still deciding which sounds better but even the fast attack didn't really get any transients.

The blurry mess happens with all compressors once you take them beyond a certain point. Dedicated mixbuss compressors reach that point much later. Mastering limiters have attack times of less than 1 micro second. I think the peakstop limiter is somwhere around 1 or 2 milliseconds. That is a huge difference in knocking down transients.

 

[FALKEN]

yes I totally believe you on that less than 1 ms. this is why I had to give up on volume and just compress it to sound good. unfortunately it came out RMS -18 and I'm not sure thats loud enough to release. I mean, lets be honest here. so my last option would be to slam it through the peakstop and just murder it, or invest in something else. everybody says aphex but those things look cheap as hell.

 

[Farview]

The other thing would be to remix. It depends on why your mixes are quiet. -18 would be loud if it was a classical guitar album, -12 is too quiet for the latest Korn release.

If it is the drum transients that are keeping you from turning it up, try compressing the drums. That will make them fatter, tame some of the transients, and make it so you can turn up the rest of the instrument around them. That will help your overall volume level.

I guess what I'm tring to say is: Your goal is to get the difference between the loudest peak and the sustained notes closer to each other. If you look at the file and you notice 1 or 2 really big peaks that are 3db above the rest of the peaks, you can zoom in and just manually turn them down to match the rest. You will have gained 3 db without harming anything. If you take that and run it through a compressor and tame the whole thing by another 2 or 3db, you have just made your mix twice as loud.

Without being able to hear it (or even knowing what style of music you are doing) I'm just shooting in the dark with some of these suggestions.