normalizing to 0.0 and -0.0 db

[Jafo]

Oh yeah... didn't even think about that. Much better argument than mine, alas, but thanks for pointing that out!

[dark_virus]

Now this is hilarious...

The best part is "it might seem trivial but its important.. "

I don't normalize anything to 0. Max is -0.1 or -0.2, according my patience setting the right number.

[mike@manike]

I never normalize, not saying you shouldn't - but I think there's better way's of going about doing what you're trying to do.

It's not in the long term going to make anything louder, if you're running something though a compressor I use the input gain instead. If you normalise something, chances are the next thing you do is turn it down a little.

I think it comes down to normalising is a destructive process, these days it doesn't matter so much because a DAWs history is more able to going back to the original sample but still, I don't think it's necessary.


Mike

[Nystul]

When listeners hear your wimpy music peaking at -0.0 (or even -0.2 if you listen to these crazy people) and then mine goes all the way to 0.0, all your fan are belong to us!

[koalaboy]

It's not that simple. Intersample peaks have been shown to be more than 6db above the digital peak in some cases.

Indeed. The only way to truly avoid intersample peaks is to.... monitor the intersample peaks

Of course, you could just normalise to something like -10dbfs and TURN THE VOLUME KNOB UP.

You should only be aiming for as close to 0dbfs (taking into account intersample peaks) during final mastering.

[flux82]

thanks all for the great laugh but seriously.. i want to normalize audio so it shows as 0.0db in a wave editor

i can't find an option to change the polarity

i have tried phase invert but it still shows -0.0db ..

come on, somebody must know?

[C00kie]

so it shows as 0.0db in a wave editor

Which one? Have you ever seen it report +0.0 as the peak value?

Ya know, you can't get a positive readout on the peak value. Ever. That would be clipping. So it's very possible your "wave editor" has the minus sign put there permanently, which makes this whole endeavour rather futile.

[Meffy]

Take jafo's advice. You may want to normalize to 0.0 dB, but you should want to normalize to a value that's actually useful. You can't squeeze a better s/n ratio out of a sound by multiplying its amplitude by a constant. The n gets multiplied along with the s.

The - should be there except for signals at the absolute maximum representable value, which are 0.0 exactly. -0.0 really means "a negative number too small to be shown with one digit past the decimal point." (For instance, -0.02 dB or -0.000071 dB) Which is good enough, except you should want -0.3 dB or so, as mentioned by other folks.

[flux82]

no, i've only seen -0.0db and 0.0db reported in wavelab and audacity and some normalizer program.

i have some audio files that hit 0.0db and some only -0.0db

is there a way to make the -0.0db into 0.0db?

the polarity suggestion seemed great but if polarity change means phase invert then it didnt work

[rifftrax]

Hmm, is that a Celsius or Fahrenheit zero?

Anywho, assuming the OP is serious, I've heard that a lot of older (and some newer) CD players can distort noticeably at -0.3 or louder. That may just be rumor, but it does kinda explain why -0.3 is pretty standard these days. (It was -3.0 in the Elder Days). I do know that running anything at its limits causes it to distort, and why would you want to spend a lot of time making something sound good, just to have it sound like crap at the final stage?

I've read this in quite a few books about mixing. I think Bob Katz recommended the exact same thing (except he said -0.2 is fine).

no, i've only seen -0.0db and 0.0db reported in wavelab and audacity and some normalizer program.

i have some audio files that hit 0.0db and some only -0.0db

is there a way to make the -0.0db into 0.0db?

the polarity suggestion seemed great but if polarity change means phase invert then it didnt work

I think you're chasing after spectres with this idea of yours that turning -0.0db into 0.0 will make all the difference. Buy a better limiter if you really want to accomplish anything at the clipping/limiting stage that will be useful. Seriously. Normalization has honestly no point whatsoever and will actually be counter-productive if you are trying to build sample libs or whatnot as it's such a waste of time.

[Meffy]

is there a way to make the -0.0db into 0.0db?

Little piece of masking tape stuck over the appropriate spot on the display.

In other words, ignore it. It is irrelevant. It means nothing; it is an illusion and a distraction placed in the way of your making music.

Instead, follow jafo's and Kim's advice and you will avoid the frustration of having to do things over again the right way.

[rifftrax]

Meffy is right.

It's funny, but it seems every new person to get into mixing or multitracking has to run into this brick-wall false concept of normalization and perfect 0.0 peak metering vs. anything else. I mean, I even remember thinking the same thing at one point like I was on a quest to normalize all my samples and have all my peaks hitting exactly 0.0 (not -0.0, not - anything), until I realized how stupid that is.

You have to get over it.

[jupiter8]

In Audacity it's called "invert" but as several here has already pointed out it's pointless.

[Meffy]

It's funny, but it seems every new person to get into mixing or multitracking has to run into this brick-wall false concept of normalization and perfect 0.0 peak metering vs. anything else.

You too?

I mean, yes, I've heard that this might be the case. Hem, yus yus.

I used to think this had to do with my being from the old analog-and-tape days, and not being used to the ways of digital audio, but apparently not.

[HTBoogie]

is there a way to make the -0.0db into 0.0db?

Little piece of masking tape stuck over the appropriate spot on the display.

In other words, ignore it. It is irrelevant. It means nothing; it is an illusion and a distraction placed in the way of your making music.

Instead, follow jafo's and Kim's advice and you will avoid the frustration of having to do things over again the right way.

I was going to suggest white out.

But, seriously, this really is an exercise in futility as far as your end product is concerned. As mentioned above, I think that this is the result of rounding errors in the normalization routines and then in the function that formats the string for display. I can appreciate the aesthetic issue here but I bet dollars to donuts that there is NO practical implication of the minus sign. Once your out to the third decimal place in dB levels NO ONE can hear the difference.

This thread did make my morning, though!