normalize question

[julien breval]

Hello,

I have finished a new track and wonder about normalization.

Everything was recorded to DP or Max/MSP in 44100 Hz / 16 bit resolution, then edited in DP. During all this time, all input or output levels were kept to 0 dB (no change). The peak level of the track is nearly 0 dB so it's still possible to normalize. The final format will be 44100 Hz / 16 bit (CD).

What is the best method for keeping the most natural sound ?

a. export to 44100 Hz / 16 bit, then normalize and export again to 44100 Hz / 16 bit
b. export to 44100 Hz / 24 bit, then normalize and export again to 44100 Hz / 16 bit
c. keep the file without normalizing as nobody will notice the difference


Best regards,
-j

[FMiguelez]

Hello,

c. keep the file without normalizing as nobody will notice the difference

This one.
Why would you want to normalize? What are you trying to achieve?

Why do you want it SO close to 0?

By a "new track" do you mean a full mix, or just an instrument?

[julien breval]

Hello,

c. keep the file without normalizing as nobody will notice the difference

This one.
Why would you want to normalize? What are you trying to achieve?

Why do you want it SO close to 0?

By a "new track" do you mean a full mix, or just an instrument?

I meant a full mix or final mix (though there's only one instrument and some realtime fx).

[FMiguelez]

.

If what you're trying to do is just more loudness, why don't you just use a compressor or a limiter? You could set the ceiling to -0.3, and set the threshold so you limit a few dBs? Just the peaks...

[julien breval]

I don't need more loudness and I want to keep the peaks.
Normalizing is only for using all the digital scale.

[Tritonemusic]

.

If what you're trying to do is just more loudness, why don't you just use a compressor or a limiter? You could set the ceiling to -0.3, and set the threshold so you limit a few dBs? Just the peaks...

Exactly.

[FMiguelez]

I don't need more loudness and I want to keep the peaks.
Normalizing is only for using all the digital scale.

Then I suppose you can normalize, but I wouldn't go higher than -0.2 anyway.

[julien breval]

Then I suppose you can normalize, but I wouldn't go higher than -0.2 anyway.

why ? is a 0dB peak somehow "dangerous" ?

[Appalachian Boy]

yes it is, many commercial cd players and such will have digital clipping on there outputs if it is set to say -0.1DB. therefore the standard is -0.3DB. just follow that rule.

[Tritonemusic]

I don't need more loudness and I want to keep the peaks.
Normalizing is only for using all the digital scale.

Sorry Julien, I posted at the same time as you did so please disregard my previous post.

[OldTimey]

just to play devils advocate here:

most commercial cd's go all the way to 0.0.

import a file from a popular cd, put it in peak and analyze it. yup they go to 0.0 all the time.

I don't think that going to 0.0 is all that dangerous, but rather how you get to 0.0.

your ceiling can be 0.3, but if you mash it up with an L3 then it's still gonna sound bad.

if you do choose to normalize, it should be the the next to final step in mastering, before dithering. chances are your mix will already be peaking @ 0.0 to 0.3 if you mastered correctly anyway, so normalization is not needed.

some old cheap crap cd players will clip at 0.0, but you probably wouldnt notice it through the crap speakers those players are pumping music through.


i never normalize anything (except vocal/speech editing i do for broadcast). it is pretty useless when dealing with music.

[gearboy]

Yeah, think of normalizing as mastering for dialog for broadcast. Back in the day when I was a recording engineer for a message and music on hold company we would RMS normalize dialog after we assembled it, sometimes from various takes. So, RMS would increase general loudness, then peak would keep everything within a certain threshold.

With music, a peak limiter, mastering EQ, and multiband compression will get you much further than Normalizing. However, these techniques introduce myriad ways to crap up your recordings if you don't know about the subtleties that you should be listening for.

[magicd]

The default ceiling setting in the MW Limiter is -.5db. The reason for this is that dither is added after the rounding and increases amplitude. The .5db provides just enough extra headroom to add dither signal and not clip.

Magic Dave

[gearboy]

Hello,
What is the best method for keeping the most natural sound ?

a. export to 44100 Hz / 16 bit, then normalize and export again to 44100 Hz / 16 bit
b. export to 44100 Hz / 24 bit, then normalize and export again to 44100 Hz / 16 bit
c. keep the file without normalizing as nobody will notice the difference

None of the above, or "C".

How many tracks do you have? If you have a computer that can handle it, convert all of that audio to 24-bit if you are using plug-ins. It'll process and sound much better. Always record in 24-bit. It sounds much better than 16 bit, has more headroom, and makes much more of a difference than higher sample rates do.

Peaks should not exceed -3dB when using Trim, though some MEs want -6dB as a safely. If you are sending this out for actual mastering, ask the ME what s/he prefers as a max peak reference. Keep everything at 24-bit.

Once you have this set up, you hopefully have a nice sounding and well balanced mix. If you are home mastering (Like I do for smaller projects), something like Ozone3 or the like with a peak limiter and dither to bring it as close to zero (I shoot for -.1dB for home mastered stuff) as possible. Dither to 16bit. Now you are ready to sequence for CD.

Check this out:

http://www.izotope.com/products/audio/ozone/

And this for some guidelines regarding home mastering.

http://www.izotope.com/products/audio/ozone/guides.html

Jeff

[Tim]

In this case where everything was recorded at 16 bit, why take it to 24 bit (unless mixing/summing through analog)? Why add dither?

If the files are going to be Mastered, I know of zero reasons to normalize, and a few reasons why not to.