I'm making my personal samples pack from my production. Usually samples aren't normalized and the volume can be really different between them. My idea is how about normalize them?

(I work in Logic X)
What are differences between Normalization on and protection clip?
What would you do if you were me?

 

No, I don't do it. I'm oldschool, and we learned early on that normalization is a destructive process.
Instead try simple and effective gain-staging. It goes a long way.

 

Just go with EBU R-128 Volume Detection normalization.
It is the best way to match different samples.

More on

Baxter is right since the early days we used to store all our media to 16bit, so turning the gain down you reduced the bit depth and then peak normalize it would bring this up, ending with even a 10 bit actual resolution.
Nowdays this is not the case since DAWs works internally up to 64bit and affect the actual resolution way less.
Just record at maximum possible bit rate at all times.

 

Normalisation is just raising the level of the sample to 0dBFS, so it also boosts the noise and everything. But it is certainly *much* less destructive than raising the level with a limiter, as at least it keeps the transients intact. However, there is no real advantages in normalising the samples these days as all the DSP processing is 32-bit or even 64-bit. I hope you know what I mean. It really doesn't make much difference so you don't need to bother with it.

Speaking of which, I really prefer to get non-limited samples in the libraries I buy or download. They can be normalised, but please don't destroy them with a limiter. That can be easily done by me if I choose to.

 

Why not ? i mean, if it's a clean sound so why not, it's not a final product, a sample is supposed to be an editable sound, you can adjust it on the prefader if it's too loud or/and on the mixing. So, in my huble opinion, you can do it if you want.

 

Nice vid. ,thnx.

 

Most likely you have different sample formats, like some 16bit and some 24bit samples.
Even volume detection normalization will do more or less harm on 16bit samples.
I would at least normalize all samples "into" 24bit resolution, if at all.

 

Of course i'm not going to use any limiter guys!

What about Normalization process on transients? Can it change and clip them? Why there's 2 tipe of normalization in logic (on or protect from overload clip), differences between them?

 

If you are making samples for your own use, do what you want as long as it sounds good to you.

Personally, I would not nomalize samples. The samplers I use have a gain function in the sample edit window, along with other standard effects further down the DSP path. Balance your sounds there, ESX24 and Alchemy both allow you processs your sampler prior to output so use your ears and spend the time to do right.

Most of the time if you normlaize that sample, you are just going to end up turning down anyways. No reason for anything to peak above 12 dbFS anyway if you are doing things right.

 

I do it anytime, it makes it much easier to see the start and endpoints of samples

 

Normalization to 0dBFS could be useful before bit-reduction, as then full dynamic range of the final bit depth is used.
NI could make a tool for bit-reduction Kontakt libraries to 16bit. This would save a ton of space without audibly affecting noise floor.
Such tool would first scan each sample for peak level, then normalized it, reduced bit-depth (with dither) and finally attenuated the respective sample in the mapping editor by its original peak value, thus preserving the original sample volume.

This could shrink every 24bit library by 25-50%, unfortunately no-one thought of it yet.