Page 4 of 62

Posted: Tue Feb 03, 2009 3:42 pm
by Mad_EP
Jalfrezi wrote:
Macc wrote:WHY normalise it though? Why do you need to?
i went to a studio once to get a tune mixed down without normalising and the engineer was saying we shouldn't have the tracks so quiet as when going out of the computer it adds noise.

does it actually cause problems when mastering? i would think they need it to be as loud as possible so that the noise level is as low as possible. not loud like brickwalling but just raising the level.

I guess it depends how you are going out of the computer... if you are playing it in real time out the CPU with shoddy A/D conversion onto tape, then some noise can be added that way, but that is an A/D issue, not normalizing issue. But if you FTP the engineer the WAV (or hand in a CD or whatever), there shouldn't be any extra noise. Most of the noise floor issues come from recording onto an analog medium and then boosting or ambient noise and/or hum from mic'ed acoustic instruments. None of those are relevant if you are doing everything ITB.

I take your point- that sure, being non-destructive, normalizing shouldn't be a problem.. but as Macc said - "Why?" The mastering engineer will probably just turn it down anyway (which is where you were to start with)...so why not just give the raw, well-mixed track?

Posted: Tue Feb 03, 2009 3:52 pm
by Sharmaji
normalizing just turns your audio up in volume, so that the highest peak hits 0db. the thing is, percieved volume isn't a function of peak level-- it's a function of the RMS. so you're not really making things louder by normalizing.

it's kind of an antiquated process that was meant to help people get every usable bit out of a 16- or 20-bit machine.

the mastering engineer was correct in a way, especially if you're sending 16 bit mixes-- you want to encode at the highest level possible so that you're not sending stuff that's only using 12 bits or so. but at 24 bits, you have a gigantic dynamic range-- the issue of too quiet pretty much goes out the window.

unless you're building samples, i can't think of a good reason to normalize.

Posted: Tue Feb 03, 2009 6:26 pm
by flateric
TeReKeTe wrote:normalizing just turns your audio up in volume, so that the highest peak hits 0db. the thing is, percieved volume isn't a function of peak level-- it's a function of the RMS. so you're not really making things louder by normalizing.
Doesn't normalising raise the rms aswell as the peak? So it kind of does make things louder?
TeReKeTe wrote: the mastering engineer was correct in a way, especially if you're sending 16 bit mixes-- you want to encode at the highest level possible so that you're not sending stuff that's only using 12 bits or so. but at 24 bits, you have a gigantic dynamic range-- the issue of too quiet pretty much goes out the window.
The engineer seems to be talking about sending signals at low levels out of the computer which would result in a higher noise floor when the level is increased (in which sense he's right) but this has little to do with sending stuff for mastering as your audio should only leave the box as data. That send I do think stuff like reverbs (as a send/100% wet) benefit from being as loud as possible but ofc the engineer may just end up replacing them.

Posted: Tue Feb 03, 2009 6:32 pm
by macc
I did write a big thing and promptly lost it :|

Don't forget that once a sample is captured, it is captured. You turn up the low-level quantisation noise with it - you aren't gaining resolution back. Try it yourself. Get a 16-bit wave file, turn it down by 60dB, save it, and then turn it up again. *

The analogue noise floor issue mentioned above is quite correct, but again that would have been recorded at the time, and playing back out via DA shouldn't add anything worrying to that - or you need to upgrade your shit!

Another issue is that a normalised file is very likely to suffer from intersample peaks. It WILL get turned down (by an ME worth his salt) before going out via DA to analogue, in order to avoid clipping the DA before it even reaches the analogue processing chain.


* For more info on bit depth, resolution, blah blah blah, see here :)

Posted: Sat Feb 07, 2009 1:21 pm
by jostov
very nice topic, learned a lot here.

Posted: Sun Feb 15, 2009 1:41 am
by wirez
DO YOUR A-B COMPARISONS AT MATCHED SUBJECTIVE LEVELS

I've actually never thought to do that and it sounds so right, that will be something I do from now on!

Posted: Sun Feb 15, 2009 1:41 am
by wirez
DO YOUR A-B COMPARISONS AT MATCHED SUBJECTIVE LEVELS

I've actually never thought to do that and it sounds so right, that will be something I do from now on!

Posted: Sun Feb 15, 2009 1:41 am
by wirez
DO YOUR A-B COMPARISONS AT MATCHED SUBJECTIVE LEVELS

I've actually never thought to do that and it sounds so right, that will be something I do from now on!

Posted: Sun Feb 15, 2009 5:21 am
by dj vision
TeReKeTe wrote:normalizing just turns your audio up in volume, so that the highest peak hits 0db. the thing is, percieved volume isn't a function of peak level-- it's a function of the RMS. so you're not really making things louder by normalizing.

it's kind of an antiquated process that was meant to help people get every usable bit out of a 16- or 20-bit machine.

the mastering engineer was correct in a way, especially if you're sending 16 bit mixes-- you want to encode at the highest level possible so that you're not sending stuff that's only using 12 bits or so. but at 24 bits, you have a gigantic dynamic range-- the issue of too quiet pretty much goes out the window.

unless you're building samples, i can't think of a good reason to normalize.
so you seriously dont normalize anything? shit

i normalize EVERYTHING! i thought i heard that you should somewhere before, i guess not? i will have to try not normalizing i guess

Posted: Sun Feb 15, 2009 9:23 am
by macc
WiREZ wrote:DO YOUR A-B COMPARISONS AT MATCHED SUBJECTIVE LEVELS

I've actually never thought to do that and it sounds so right, that will be something I do from now on!
That actually was worth saying three times :D

Posted: Sun Feb 15, 2009 12:50 pm
by wirez
Macc wrote:
WiREZ wrote:DO YOUR A-B COMPARISONS AT MATCHED SUBJECTIVE LEVELS

I've actually never thought to do that and it sounds so right, that will be something I do from now on!
That actually was worth saying three times :D
Lmao, something was going crazy with the submit button last night, I did that a couple of times :\

Posted: Sun Feb 15, 2009 1:46 pm
by macc
:lol:

Usually people take the piss but I was serious - that really was worth repeating! :D

Posted: Sun Feb 15, 2009 5:51 pm
by Sharmaji
DJ Vision wrote:
TeReKeTe wrote:normalizing just turns your audio up in volume, so that the highest peak hits 0db. the thing is, percieved volume isn't a function of peak level-- it's a function of the RMS. so you're not really making things louder by normalizing.

it's kind of an antiquated process that was meant to help people get every usable bit out of a 16- or 20-bit machine.

the mastering engineer was correct in a way, especially if you're sending 16 bit mixes-- you want to encode at the highest level possible so that you're not sending stuff that's only using 12 bits or so. but at 24 bits, you have a gigantic dynamic range-- the issue of too quiet pretty much goes out the window.

unless you're building samples, i can't think of a good reason to normalize.
so you seriously dont normalize anything? shit

i normalize EVERYTHING! i thought i heard that you should somewhere before, i guess not? i will have to try not normalizing i guess
I used to, for samples, when i was using an MPC. but nowadays? Never, because it's not a musical solution to anything.

Posted: Sun Feb 15, 2009 6:04 pm
by r
normalizing is quite cool if you work lot with vocals and stuff you recorded yourself. Normalize the words and the stuff. Dont normalize the breath takes or whatever that isnt a word. Itll get more a poppy edge because of the 'compression' kinda feeling. Every word is bring to 0db. This is a 'better' way to threat vocals instead of compressing them cause otherwise the breath takes get the same volum as the spoken/singen things. Normalize gives a cetrain 'edge'. The bit distorted 'edge'.

Bout the topic start. Search for balance. there's no rule about how much db or something like that. Everything is about balance.

Posted: Sun Feb 15, 2009 6:47 pm
by emef
this is a great thread, i'm taking a lot of advice from these posts
i find myself reading a lot of macc posts here and on doa
so, cheers macc

Posted: Sun Feb 15, 2009 7:59 pm
by fatasfunk
This is a great thread. Speaking as a fellow mastering engineer, Maccs advice is absolutely spot-on. I always sigh when i'm sent a track with Fuck-all headroom! One thing i'll add is ALWAYS mix down at 24 bit or higher if sending to an ME as the issue of degradation is minimised and the doubters can mix quieter with no fear ;)

Posted: Sun Feb 15, 2009 9:35 pm
by metalboxproducts
Great thread

Posted: Sun Feb 15, 2009 10:01 pm
by macc
emef wrote:this is a great thread, i'm taking a lot of advice from these posts
i find myself reading a lot of macc posts here and on doa
so, cheers macc
That's cos I've got a big fuckin mouth :lol:

I dunno... having been through a squillion headaches myself, it just makes sense to help other people out if I can, save them the fuckin about.

Mind you, I am on holiday for a week so you have a week off from my waffling :6:

Posted: Fri Feb 20, 2009 11:06 am
by daft cunt
This thread changed my mixing skills!!

Thanks a lot Macc!

Posted: Fri Feb 20, 2009 11:30 am
by serox
yeh macc has helped me out loads.

Its best to take advice from people who clearly know more than you about science/physics lol