Normalizing?

[soldierone]

I've heard a lot of different thoughts on using a normalizer in the mastering process... some people really like using it to get even levels on their tracks and a lot of people warn that you should stay away from it like the plague. So I'm looking for your thoughts on this particular process.

Recently I had the chance to check out Wavelab over at a friends house. He showed me some of the features of the program including the meta normalizer, which he used to make the average rms the same across the board on a handful of tracks. I liked the overall result, but was wondering where in the mastering chain you might use such a process.

In the past I've done some basic matering of my own on some of my releases, and I pretty much stuck to doing things in this order: basic editing, (including removing excess silence from the beginnings and ends of tracks,) EQing, adding reverb as needed (I would use this on certain tracks that I wanted to change the character on, but overall this would be something I usually don't mess with,) adding compression and limiting and then I would use a dither to convert my 24bit files down to 16bit.

If I were to use a normalizer in the manner that I described above, where would I use it in my existing process to get the best results from it?

 

[Massive Master]

Standard normalization brings the peak volume to a particular point (pretty useless). WL's MetaNormalizer takes a guess at the overall level (equally useless in a different direction).

Wouldn't you rather just use your ears?

 

[soldierone]

Standard normalization brings the peak volume to a particular point (pretty useless). WL's MetaNormalizer takes a guess at the overall level (equally useless in a different direction).

Wouldn't you rather just use your ears?

Historically that's exactly what I've done. I've never actually used normalization, so maybe that's something playing in my favor?

 

[Massive Master]

I'd say so.

 

[mixsit]

needless redundant processing based on a mathimatical 'norm-

... He showed me some of the features of the program including the meta normalizer, which he used to make the average rms the same across the board on a handful of tracks. I liked the overall result, but was wondering where in the mastering chain you might use such a process.

In the past I've done some basic matering of my own on some of my releases, and I pretty much stuck to doing things in this order.. snip ..adding compression and limiting and then I would use a dither to convert my 24bit files down to 16bit..

Basically the way I see it is here you (we ) go to the trouble to set up and tweak all these finalizing' tools - gain, density, relative peak, relative song loudness and impact, their roll in the play list. Let's say you do use it in there somewhere.
-The tools are all in place, Abnormalizing rolleyes:) nine times out of ten still only gets you 90% there..
...It's back to the those same gain tools to dial it in the last gnat's ass.

 

[.guitar]

There's a lot of misinformation about how bad normalizing is.

If you do it more than once, then yes, it is bad, and you should never have to.

If you have everything in the song done, and are happy with how it sounds, have the compression right, etc., then normalizing right before dithering is fine. Some people say to use a limiter instead (like Waves L1/L2).

I normalize to -.5 to -1db right before dithering. Whatever level you choose is up to you. But I personally never go all the way up to 0db.

With that said: Never normalize individual tracks, unless it's non-destructive...and even then, just use the faders.

 

[RAMI]

There's a lot of misinformation about how bad normalizing is.

And your post shows that there's even worse misinformation on how good it is.

 

[SouthSIDE Glen]

It sounds like what the OP is talking about is RMS normalization, not your run-of-the-mill peak normalization. The problem with RMS normalization (a.k.a. "volume balancing") is that there's no guarantee that any two tracks are going to have the same perceived volume just because they have the same average RMS volume, or that you really *want* two tracks to have the same perceived volume.

If you throw a sonicly fairly sparse ballad into a playlist with headbanging wall of distortion, are you really going to want or expect them both to sound optimum at the same RMS, regardless of what that RMS level actually winds up being?

Even if you picked a middle compromise level of, say, -13dbRMS, most headbangers would be disappointed that the anthem isn't smashed enough and there's a good chance that the ballad might be pushed just a tad too far.

To make a point of it, think of it in terms of a direct answer to your question: RMS normalization would be used in place of your compression/limiting stage, since that is basically what the normalizer is doing on it's own. The difference is the computer's making it's dynamics adjustments completely impartially, based on an arbitrary set of numbers instead of what things actually sound like to the human ear.

My personal advice would be that if you stick on your current course and go without the normalizer (and also don't be afraid to break both the EQ and the dynamics parts of your process into smaller and interleaved bites), that your results will wind up sounding much better and - just what everybody seems to want these days - more professional-sounding than if you left the decisions to a PC.

G.

 

[.guitar]

And your post shows that there's even worse misinformation on how good it is.

Coming from someone who probably just repeats what others say without any actual knowledge of how bit-depth works and what actually happens during normalization or how floating point algorithms work and probably no computer programming background either.

But yeah...

 

[.guitar]

It sounds like what the OP is talking about is RMS normalization, not your run-of-the-mill peak normalization.

If that's the case, then yeah, it's bad.

But a single, one-time normalization (not RMS normalization) before dithering is not harmful if no more processing will be done on the audio.

 

[SouthSIDE Glen]

But a single, one-time normalization (not RMS normalization) before dithering is not harmful if no more processing will be done on the audio.

Agreed. All peak normalization is at that point is is a volume control raising the overall gain until the highest peak reaches maximum level.

No harm in that *as long as we're talking about mastering one song and one song only*. But if it's part of mastering one of a series of tracks in an album, then it makes nor more sense than RMS mastering, because one has to - or at least should - take the surrounding material in consideration and not just normalize the track in isolation.

G.

 

[mixsit]

There's a lot of misinformation about how bad normalizing is.

To our credit, no one here at least said it's bad. On average it's be quite a bit like making similar processing by other means.

If you do it more than once, then yes, it is bad, and you should never have to.

I would ask what is bad about one more gain change out of a string of several along the way...

If you have everything in the song done, and are happy with how it sounds, have the compression right, etc., then normalizing right before dithering is fine..

But then you sort of made the case of my point earlier..
If you're doing gain at the comp/limiter -then Anormalizing is doin it twice.

 

[.guitar]

To our credit, no one here at least said it's bad.

I've been a member here for awhile and it seems every time the topic of normalizing comes up, in-comes the responses to "NEVER normalize", "normalizing will make it sound shitty", etc.

I would ask what is bad about one more gain change out of a string of several along the way...

Destructive editing should be kept to a minimum. Obviously, a few gain changes aren't going to make any noticeable difference.

But then you sort of made the case of my point earlier..
If you're doing gain at the comp/limiter -then Anormalizing is doin it twice.

Yes, there is no reason to normalize if you adjusted to the right gain during compression or limiting.

 

[mixsit]

I've been a member here for awhile and it seems every time the topic of normalizing comes up, in-comes the responses to "NEVER normalize", "normalizing will make it sound shitty", etc. .

Yeah it's true, a lot of over stating goes on, in the land of zero dynamics and Smashtering, the ever popular 'you'll bring your noise level up.. Yep.

I don't say never.

I just ask why.

 

[RAMI]

Coming from someone who probably just repeats what others say without any actual knowledge of how bit-depth works and what actually happens during normalization or how floating point algorithms work and probably no computer programming background either.

But yeah...

Blah blah blah.................................................
You display more ignorance every time you post. Keep going.






But yeah.............Not.

 

[.guitar]

Blah blah blah.................................................
You display more ignorance every time you post. Keep going.






But yeah.............Not.

You don't even know what you're talking about and you call me ignorant.

k

 

[DAS19]

Ive read normalization is unnecessary and doesn't actually make any difference.

 

[SouthSIDE Glen]

Ive read normalization is unnecessary and doesn't actually make any difference.

Peak normalization is nothing more than a volume change. There's nothing that peak normalization does that a fader or volume control won't do. The only "special thing" that peak normalization does is that it automatically looks at the whole track and determines it's peak volume level, and the increases the overall volume of the track by as much as possible before clipping the top pf the biggest peak.

In that way there's nothing "wrong" or "dangerous" about peak normalization any more than there's anything wrong or dangerous about pushing a track fader by that same amount. It's just a gain control; nothing more, nothing less. Like any tool in any profession, the only time it's "bad" is when it's misused. Where the problem comes in is in *where* and *why* people mistakenly use it.

In this case, when people use it on individual tracks in a mix before mixing, all they are doing is stealing mix headroom from themselves and they'll just have to turn everything back down again anyway; it's just messing with normal gain structure. In that way it usually should not be used on individual mix tracks.

And when people use it on final mixdowns, it's fine as long as that's the only song whey have to worry about it. it may not give one the maximum optimal volume - i.e. it may not be the absolute best solution - but it certainly won't do anything damaging to use it in that way. Where it becomes a problem here is when one is mastering a collections of songs as in a demo, EP or CD. Many newbs believe that "normalizing" means making everything the same volume, but it doesn't. There's is much more involved than simply lining up the maximum peaks of individual tracks if one wants to make the tracks sound like they are playing back at the same volume. So in that light, peak normalization is the wrong tool for the intended job.

Now what the OP is talking about (I think) is not typical peak normalization, but rather RMS normalization. This does sound better than peak normalization to the OP because it does do a somewhat better job than peak normalization at making two songs sound of similar volume. But it's still a way over-simplified approach to the problem of mastering for equal playback volume that simply cannot do the job that the human ear can and will leave most mastering jobs lacking on that score.

G.

 

[RAMI]

But it's still a way over-simplified approach to the problem of mastering for equal playback volume that simply cannot do the job that the human ear can and will leave most mastering jobs lacking on that score.

G.

Sorry Glen I hate to bring you into these little disagreements, but I have to quote you here because that was pretty much what I was getting at.

Also, if you have one peak that stands out disproportionately, then normalization is not going to help, it will hurt.

I'm done with arguing with an idiot who has never had anything intelligent to say, and who has very often been exposed as not knowing what he's talking about. Now, this Bozo is telling me what he THINKS I don't know about computers blah blah blah....which only shows, once again, that he should keep his mouth shut more often if he doesn't want to come off like the idiot he is.

I'll leave this thread in the hopes it stays on track and let Bozo go on about algorithms and bit-depths. Talk about repeating shit that you don't understand, that was a classic.

I am now un-subscribed to this thread. Go nuts.

 

[SouthSIDE Glen]

Sorry Glen I hate to bring you into these little disagreements, but I have to quote you here because that was pretty much what I was getting at.

You're not dragging me into anything, Rami. I do find it a bit of a shame that you and .guitar (nee danny.guitar) don't get along, though, because I actually like both of you guys.

I don't know whether it's the holidays (I know, I know, you're in Canada and Thanksgiving doesn't quite count) or the economy or the lack of decent ja, but it seems like there's a bunch of us that are a little more on edge than usual these days. I have not been immune to it either, I admit. Maybe we could all stand to step back and take a deep breath for a minute. I'm just thankful that my gas station dropped their price on BP regular to $1.67/gal when I filled up yesterday. That's down a full $3.00/gal from July! Woo hoo!

Also, if you have one peak that stands out disproportionately, then normalization is not going to help, it will hurt.

I'm not sure I follow that one, RAMI. It certainly won't help with anything, but in and of itself I don't see it doing any actual harm.

Let's say your average peak level is somewhere around -6dBFS (just for example), and you have one renegade peak of -2dBFS. Peak normalization won't do anything other than boost the gain on the track by 2dB. Assuming we're talking about applying it to the mixdown, no real advantage to doing it, but no real harm in doing that either.

And if we're talking about RMS normalization, then one or two renegade peaks will have very little effect on the RMS calculation.

(Of course, if my mixdown has one or two renegade peaks, I'll manually edit them down before I do *any* further mastering, but that's another thread altogether )

Or am I misunderstanding what you're saying?

G.