Peaks, Normalization, yadada....

[Seafroggys]

Hey guys, so we're just about done mixing an album, and obviously because there's no money involved, I am going to do "self-mastering". Obviously very basic stuff, minor two-buss processing, set things up in CD Architect, and done.

What I plan to do for workflow is to set up all the stereo mixes in the same timeline, get them the same loudness, run my faux mastering chain (probably BaxterEQ, TesslaPro, and FerricTDS) on the master so the sounds are unified, with potentially individual track stuff as well, but unlikely as I got the mixes to sound pretty damn good on their own.

Now with all the tracks being the same volume, when I normalize them before dithering to 16-bit, basically they're at the mercy of whichever song has the loudest peak. I'm not fighting any loudness wars here, but at the same time I don't want to lose out on extra volume for everything just because one snare hit on track #5 was 3 db louder.

In the 80's and early 90's, on most CD's, weren't there usually one peak on one track anyway? Not every track necessarily hit -0.1? Is this an acceptable practice, or was that just a product of the technology of the times because look-ahead limiters didn't exist? Would it be smarter for me to "master" the stereo tracks individually and get them to the same RMS by using meters? I kinda want to do the former for a more organic sound. For those sticky snare hits, would you go back into the mix and volume automate that snare hit? Usually when I've done automation on individual percussion hits it sounds unnatural. Would a high ratio compressor work better on just that one snare hit be more organic?

This is more a discussion of what others do, rather than me looking for "the answer." But I'm curious on what others think.

 

[bouldersoundguy]

After evening out the differences use a mastering limiter like Wave Hammer to get the level where you want it, taking into consideration that there's a trade off between volume and dynamics. If you do use Wave Hammer go to the Volume Maximizer tab and set the Output to -0.3dB, then slide the Threshold down until you get the volume you want. Try different Release settings. I usually bypass the compressor but you might find it useful.

Other mastering limiters may work as well or better depending on the material and your preferences. The nice thing about good mastering limiters is that you generally just adjust the threshold and the limiter takes care of the make up gain. That's much easier than using a general purpose compressor/limiter.

 

[GONZO-X]

you go for volume during mastering, not mixing.
don't mess with normalization and all that stuff....


keep your mixes down to peaks of about -10, good clean sonics.


go for your 'commercial levels' during mastering.

if you do not do this, you will end up with gritty sonics, tiring mixes that no one will want to listen to more than once.

 

[Seafroggys]

^^^ .....I hate to sound like a jerk, but.....you can read, right?

Other mastering limiters may work as well or better depending on the material and your preferences. The nice thing about good mastering limiters is that you generally just adjust the threshold and the limiter takes care of the make up gain. That's much easier than using a general purpose compressor/limiter.

I want to try and avoid a brickwall, otherwise I wouldn't have asked this question, as a brickwall is a no-brainer solution to the issue I'm having. Pretty much any limiter plug-in I have used does auto-makeup gain, or at least has the option. Regardless, it is a faster way than figuring which peaks are maxing out and going into the mixes and fixing those.

 

[sjjohnston]

Thoughts:

First the obvious: if you've got all the tracks just the way you want them to sound in relation to one another, and there's one song with a peak that's 3 dB louder than anything else in any other song, you have two choices:
- Leave it so the peaks of all the songs except the one are 3 dB (or more) below full scale, or
- Knock that peak down by 3 dB
(Or something in between)
If you want to knock down the peak, I'd probably just use a limiter on it. Of course, you could put in a small, quick fader move, since you're still mixing. I suppose that would be tidier, at least so long as it doesn't inspire you then to put a little move on the next loudest snare hit, then another after that, etc. etc. Typically (it seems to me) a peak isn't the result of a single snare hit that's significantly louder than all the other snare hits, but a snare hit that happens to align with a loud word and more pronounced attack on a guitar chord or some other serendipitous combination of things that pokes up a little higher. A high ratio compressor is approximately the same thing as a limiter.

Then an observation: unless you're fighting a loudness war, 3 dB doesn't mean much.

And some notes on your other questions:

In the 80's and early 90's, on most CD's, weren't there usually one peak on one track anyway? Not every track necessarily hit -0.1?

Hard to say anything enormously general, and when you say "early 90's" you're already getting into the limit-everything-so-it-looks-like-a-rectangle era. But in the '80s it would not be unusual to have CDs which never hit -0.1 dBFS even once. I'm looking at a (very) commercial '80s CD, and the peak on one track is -6.3 dBFS.

Is this an acceptable practice or was that just a product of the technology of the times because look-ahead limiters didn't exist?

I'm not sure that's a one-or-the-other question. I suppose it was a product of the technology of the times ... but, at least in my view, it was and remains an acceptable practice. Were people sitting around in the '80s saying, "Boy, I wish this CD were louder"?

Would it be smarter for me to "master" the stereo tracks individually and get them to the same RMS by using meters?

There's nothing wrong with measuring RMS as an aid to matching volume between tracks, but: (i) ultimately, it's about the volume as perceived by the listener, which may depart from RMS and (ii) you are allowed to have some tracks be louder or quiter than others.

Finally, a bottom line, per me:
I'd just get all the tracks to sound the way you want in relation to one another, and if that means one track peaks at -0.1 dBFS and another at -4 dBFS don't worry about it.

 

[GONZO-X]

^^^ .....I hate to sound like a jerk, but.....you can read, right?



I want to try and avoid a brickwall, otherwise I wouldn't have asked this question, as a brickwall is a no-brainer solution to the issue I'm having. Pretty much any limiter plug-in I have used does auto-makeup gain, or at least has the option. Regardless, it is a faster way than figuring which peaks are maxing out and going into the mixes and fixing those.

WHOA, BUDDY..
don't be a jerk.

trying to help here, it's obvious you are looking for answer, eh?

it takes skill to use a brickwall limiter the way it was intended to be used.
otherwise, anybody could do it, and everybody's mastered home mixes would sound just great!

'cept they don't.

seriously, i don't know why i bother trying when i get hit by crummy vibes....

 

[Seafroggys]

WHOA, BUDDY..
don't be a jerk.

I know I came across as one, and I'm sure you were trying to be helpful, but when I saw your post, it looked like you didn't even read mine at all. You weren't even talking about anything I was addressing in my question/statement. You go on about how normalization was not for mixing, but mastering, when my question was posted in the Mastering sub-forum and specifically addresses the mastering process. Then you go about recording levels when I'm well beyond that stage in the recording process....not to mention I generally record at -12 to -18, depending on the source.

So that's why I came across edgy because it seemed to me that even if you did bother to read what I posted, you only saw certain key words and went off on those words, even though your content had nothing to do with what I said. To be honest, I was kinda annoyed by that.

If you want to knock down the peak, I'd probably just use a limiter on it. Of course, you could put in a small, quick fader move, since you're still mixing. I suppose that would be tidier, at least so long as it doesn't inspire you then to put a little move on the next loudest snare hit, then another after that, etc. etc. Typically (it seems to me) a peak isn't the result of a single snare hit that's significantly louder than all the other snare hits, but a snare hit that happens to align with a loud word and more pronounced attack on a guitar chord or some other serendipitous combination of things that pokes up a little higher. A high ratio compressor is approximately the same thing as a limiter.

I'm thinking more and more this is the case. My question is more of a technical one than an artistic one. I've seen very dynamic albums that peak at least once every track. I see what you're saying about peaks though, its a combination of many factors, and I'm definitely with you on that. I don't think it would be the end of the world if I applied a limiter that caught 8 peaks on an 11 song album, so I can get a dB or two without compromising anything.

Just to test, since I'm away from my studio, I put the raw mixes into Audacity and applied a primitive brute force limiter. As it is, the DR is 16, and I managed to squish it down to 8 before I noticed distortion on my PC speakers, and got it down to 4 when the distortion became overly obnoxious. So again, I don't really have much to worry about if I'm slicing off a peak here and there, when 99% of the song will be untouched by the limiter. Thanks for the insight!

 

[bouldersoundguy]

Although I have thought about it a lot I find experimentation and listening more expedient than worrying about it too much beforehand. If I were you I'd try a good limiter and let your ears decide if you need to go back and tweak at the mix level.

 

[mixsit]

I do whatever mastering' in Sonar and I'll just toss in a few thoughts, one is an option, line your tracks up in sequence on the time lime, but each one it's own track. Possible advantages- can be easier for unique treatment on a per-song basis.
I know there's track and/or clip effects and other progs that make this completely unnecessary, so it depends.
As sjjhonston mentioned ears to set relative song loudness
Final compression to do all the normal things you might want it for for but also see it and/or the limiter in how they effect density. Final loudness, and the effect of the compression and/or the limiter then can be three independent variables.
The limiter for example can be set " -.5 " globally (a fairly typical placement') but neither it nor the compressor have to be global. Ie it can be that a song sounds 'better driven into either (or both) of them, w/o necessarily wanting to be setting up at the top of output ceiling.
In either case 'normalizing isn't needed, not even in the running. You set the final 'loudness.
And that final loudness is set after you dial in the density ..and style'

 

[GONZO-X]

I know I came across as one, and I'm sure you were trying to be helpful, but when I saw your post, it looked like you didn't even read mine at all. You weren't even talking about anything I was addressing in my question/statement. You go on about how normalization was not for mixing, but mastering, when my question was posted in the Mastering sub-forum and specifically addresses the mastering process. Then you go about recording levels when I'm well beyond that stage in the recording process....not to mention I generally record at -12 to -18, depending on the source.

So that's why I came across edgy because it seemed to me that even if you did bother to read what I posted, you only saw certain key words and went off on those words, even though your content had nothing to do with what I said. To be honest, I was kinda annoyed by that.



I'm thinking more and more this is the case. My question is more of a technical one than an artistic one. I've seen very dynamic albums that peak at least once every track. I see what you're saying about peaks though, its a combination of many factors, and I'm definitely with you on that. I don't think it would be the end of the world if I applied a limiter that caught 8 peaks on an 11 song album, so I can get a dB or two without compromising anything.

Just to test, since I'm away from my studio, I put the raw mixes into Audacity and applied a primitive brute force limiter. As it is, the DR is 16, and I managed to squish it down to 8 before I noticed distortion on my PC speakers, and got it down to 4 when the distortion became overly obnoxious. So again, I don't really have much to worry about if I'm slicing off a peak here and there, when 99% of the song will be untouched by the limiter. Thanks for the insight!

your issue really originates in the mixing process.

it's all connected.

you don't want to have to fix something with MBC at the mastering stage, when it can be addressed in the mixing stage.
that's working backwards.

also, the art of 'mastering' is not looking at the math.
it's listening with the ears.

whether one song peaks at -0.1 or not, really has no bearing on the 'art' of making a group of songs flow together.

this is a purely subjective judgement, and when you throw caution to the wind and simply run the group thru a normalization process, you are doing yourself no favors.

 

[Waltz Mastering]

In the 80's and early 90's, on most CD's, weren't there usually one peak on one track anyway?

Not really. Early cd masters would have headroom throughout, and many cd's would be a couple dB's away from digital full scale. This progressively changed through the mid to late 90's when br limiters and clipping became vogue.

Not every track necessarily hit -0.1?

Is this an acceptable practice, or was that just a product of the technology of the times because look-ahead limiters didn't exist?

If you're not using a br limiter I wouldn't be concerned with -.1dBfs. Anywhere below 0 dBfs will suffice. Many ME's will top out below -.3 or so to help prevent the audible effects of intersample peaks when the lossless files are converted to a lossy format.

Would it be smarter for me to "master" the stereo tracks individually

Yes. I like working that way

and get them to the same RMS by using meters?

No. Rms can be deceiving when it comes to perceived loudness. Use your ears and listen for continuity through vocal level and the frequency balance of the whole track as you finish and compare them to each other back to back, .. or compare the chorus's, where most of the energy is..

I kinda want to do the former for a more organic sound. For those sticky snare hits, would you go back into the mix and volume automate that snare hit?

Yes. Volume/fader automation is always easiest when taken care of in the mix stage since you have the sesions there.

Would a high ratio compressor work better on just that one snare hit be more organic?

Fader automation in the mix usually does the trick. ..but I would use the compressor on the whole track if it needs dynamic control. If you want to forgo a br limiter, then a compressor with a medium attack and quick to med release can work well. Again just listen that you aren't cutting into the leading edge or transient detail of the track to much. Low ratios and not much gain reduction is also a safe bet.

This is more a discussion of what others do, rather than me looking for "the answer." But I'm curious on what others think.

Mastering techniques have changed through the decades and not trying to even come close to loud is noble these days, ..it's also a bit more challenging without the aid of a br limiter because the perceived levels from song to song become a bit harder to judge or match. Ime, normalization is rarely if ever used in mastering.

There's several current artist that I listen to who's album are mastered in the way you describe (no limiters). One being John Frusciante's solo albums, ..ex gtr player for the Chili Peppers. His albums are dynamic. gl

 

[GONZO-X]

waltz,
you said the same thing i did.
thanks.

 

[Seafroggys]

Good info Waltz, definitely. The RMS thing is something I figured out earlier this week, I did a mixdown of all the songs and put them on a single timeline and adjusted the faders by looking at a VU meter (easier to follow than RMS meters, methinks), and even then, some songs were definitely different volumes, but it was mostly consistent. The mixes sound great, I'm gonna do the final "mastering" stage tonight with the singer and its gonna be pretty killer. I'll take some of your words into consideration as well, and won't worry too much about the peaks. I'll just make it sound good, even if there is some unused headroom.

 

[Waltz Mastering]

waltz,
you said the same thing i did.
thanks.

Yeah, I was just going to quote you, but it seemed like taking the easy way out

 

[Bobbsy]

Don't automatically avoid the brick wall EVERY time. Use your ears and common sense.

Certainly on the mixes I do it's quite common to have one or two very brief transients that, if left unprocessed, can keep the whole mix six or eight dB lower than it needs to be. The thing with transients is that you often don't even notice them when listening--or at least they are so brief you don't really hear them as loud. In that sort of case, a hard limiter with the parameters set by you by ear (and a quick look at where things sit on the waveform) can buy you a lot of extra level without affecting 99.99% of the dynamics.

What you don't want to do is use a hard limiter to destroy every peak and every inch (or dB) of dynamic range. Hard limiting is a blunt instrument that, if used, you need to keep under control.

As a general comment, I'm always a bit concerned at blanket statements of "never use xxx effect". Things like normalising and hard limiting are just tools like any other and all have their places--so long as you're careful about where they're used and how you set the parameters.

 

[Mo Facta]

If you've got a Steinberg dongle, you can download the Wavelab 8 trial and use the Meta Normalizer in the audio montage to get your tracks in the same RMS ballpark. If you need to edit one snare down, you can do that as well.

And look, brickwall limiters get a lot of flack by some but fact of the matter is that they can work well if you don't slam them. I sometimes use the Waves L1 at the track level to limit peaks and find it quite transparent. I also use the L2 and L3 for mastering and if you don't kill more than 3dB (the manual says 6dB) you shouldn't notice a huge difference from the original balance if any at all.

May the force be with you.

Cheers

 

[Seafroggys]

Thanks guys, the finished songs sound pretty killer. I ended up normalizing the entire album, and put on a brickwall at -2 dB and autogained to -.3 dB. It maybe affected 20 peaks in the whole thing, and nothing was affected more than 1 dB or so. So its pretty transparent. When we got the album artwork all done and the digital copies posted, I'll make sure to post it on here.

 

[ashcat_lt]

Don't automatically avoid the brick wall EVERY time. Use your ears and common sense.

Certainly on the mixes I do it's quite common to have one or two very brief transients that, if left unprocessed, can keep the whole mix six or eight dB lower than it needs to be. The thing with transients is that you often don't even notice them when listening--or at least they are so brief you don't really hear them as loud. In that sort of case, a hard limiter with the parameters set by you by ear (and a quick look at where things sit on the waveform) can buy you a lot of extra level without affecting 99.99% of the dynamics.

What you don't want to do is use a hard limiter to destroy every peak and every inch (or dB) of dynamic range. Hard limiting is a blunt instrument that, if used, you need to keep under control.

As a general comment, I'm always a bit concerned at blanket statements of "never use xxx effect". Things like normalising and hard limiting are just tools like any other and all have their places--so long as you're careful about where they're used and how you set the parameters.

This.

The one or two big spikes are not the transients that give the mix its impact. I call them abberant peaks. They are only there because a number of tracks just accidentally happen to be pushing in the same direction at the same time. They are almost always really fast - the first half cycle of the wave, just a few samples sometimes. You can squash those down to sit nicely with the more reasonable peaks and nobody will know the difference, but it can allow you to more than double the overall volume of the mix. I don't have any idea how you would automate something like that out in the mixing stage since it's not just "one snare hit" but rather a snare hit and a bass note and a guitar chord and...

A problem I've been having lately is that these little fuckers get through my limiters! There is very little as frustrating as setting up a limiter at say -3db and still seeing red on the peak meter! OK, so the attack time is too slow? But it starts to click and clack and sound like shit if I turn it down. One of my plugins just stops working altogether if the attack time is too fast!

But actually, I'm not even completely convinced that it's the attack time that does it. I've tried using a limiter to keep a positive-feedbacking delay at a reasonable level, but even that relatively steady-state material will just push right through most of the limiter plugs that I have. Pisses me right off!

 

[bouldersoundguy]

A problem I've been having lately is that these little fuckers get through my limiters! There is very little as frustrating as setting up a limiter at say -3db and still seeing red on the peak meter! OK, so the attack time is too slow? But it starts to click and clack and sound like shit if I turn it down. One of my plugins just stops working altogether if the attack time is too fast!

But actually, I'm not even completely convinced that it's the attack time that does it. I've tried using a limiter to keep a positive-feedbacking delay at a reasonable level, but even that relatively steady-state material will just push right through most of the limiter plugs that I have. Pisses me right off!

What limiter are you using? The mastering limiter plugins I've used are absolute barriers to peaks. I've had trouble using general purpose compressor plugins as limiters.

 

[Seafroggys]

That's what it sounds like to me. I use the Classic Master Limiter for years, and I've never had anything pass through its threshold.