Normalizing (mastering)
Normalizing (mastering)
Hi, I'm a french user of ableton live and I've got some issues when I render/normalize/"masterise" my tracks.
The thing I usually do to get a "good" render (I mean loud volume, that sounds like commercial records) is:
-I first render my track with normalize option ON (and 24bit depth)
-I create a new project with my rendered wav file, and I put the volume of the track to +6db (and I leave the master to 0db)
-I render this new project without normalizing (24bit depth again, but that's not the point here, I think)
This usually work well, but I've got 2 questions:
1/Is there a better way (but not too complicated) to do this? I mean to get a good volume, not to do a real mastering work.
2/I had some problems with voices and sax (medium/high frequencies) that seems to saturate when I render my song the second time (+6db), so I guess that theses instruments are going over the 0db limit during the first rendering (which is normalized), how's that possible?
Although I have no problems when the louder instrument is the drum kick or the bass.
Please answer quick if possible.
Thanks very much.
Flow'
The thing I usually do to get a "good" render (I mean loud volume, that sounds like commercial records) is:
-I first render my track with normalize option ON (and 24bit depth)
-I create a new project with my rendered wav file, and I put the volume of the track to +6db (and I leave the master to 0db)
-I render this new project without normalizing (24bit depth again, but that's not the point here, I think)
This usually work well, but I've got 2 questions:
1/Is there a better way (but not too complicated) to do this? I mean to get a good volume, not to do a real mastering work.
2/I had some problems with voices and sax (medium/high frequencies) that seems to saturate when I render my song the second time (+6db), so I guess that theses instruments are going over the 0db limit during the first rendering (which is normalized), how's that possible?
Although I have no problems when the louder instrument is the drum kick or the bass.
Please answer quick if possible.
Thanks very much.
Flow'
Oh stop doing what you are doing right now.
When you go over 0dB, thats clipping (i.e. digitial distortion) and its not a good type of distortion. And since you normalize, then boost 6dB more, you are clipping at least some of your song.
What part of normalization don't you understand here? Normalizing finds the maximum amplitude in your song, subtracts that value from the Max peak possible (0dBfs), and adds that difference to your entire track. if your song already goes over 0, Normalizing does nothing.
So your first render will hit 0dB, and when you reimport it and boost it +6dB, you are now going 6dB over the max. All the values that go over 0 are set to 0... hence I say you will get square waves.
And just because the master fader is set to "0" doesn't mean your song is set to 0dBfs!!... it just means the master fader is neither boosting or cutting your signal.
A better, easy way is to use a compressor. This can boost the average volume up to the level you seem to like, without giving you square waves all over the place. Drop some of the Compressor presets onto your master track and see what they do to your sound. And read the manual for what all those settings are for. Then read tarakeith's guide to mastering posted here on the forums.
When you go over 0dB, thats clipping (i.e. digitial distortion) and its not a good type of distortion. And since you normalize, then boost 6dB more, you are clipping at least some of your song.
What part of normalization don't you understand here? Normalizing finds the maximum amplitude in your song, subtracts that value from the Max peak possible (0dBfs), and adds that difference to your entire track. if your song already goes over 0, Normalizing does nothing.
So your first render will hit 0dB, and when you reimport it and boost it +6dB, you are now going 6dB over the max. All the values that go over 0 are set to 0... hence I say you will get square waves.
And just because the master fader is set to "0" doesn't mean your song is set to 0dBfs!!... it just means the master fader is neither boosting or cutting your signal.
A better, easy way is to use a compressor. This can boost the average volume up to the level you seem to like, without giving you square waves all over the place. Drop some of the Compressor presets onto your master track and see what they do to your sound. And read the manual for what all those settings are for. Then read tarakeith's guide to mastering posted here on the forums.
jesus christ
dont ever do that again
EDUCATE YOURSELF
dont ever do that again
EDUCATE YOURSELF
Anything at my disposal..
http://soundcloud.com/jagle
http://soundcloud.com/jagle
-
solacerodgers
- Posts: 191
- Joined: Wed May 21, 2008 9:35 pm
- Location: myrtle beach
- Contact:
i just dont get how someone with an ounce of sense didnt know this is a
TERRIBLE TERRIBLE IDEA
ya just throw the fader up
itll be techno great
TERRIBLE TERRIBLE IDEA
ya just throw the fader up
itll be techno great
Anything at my disposal..
http://soundcloud.com/jagle
http://soundcloud.com/jagle
First of all, thanks for your answers.laird wrote:Oh stop doing what you are doing right now.
When you go over 0dB, thats clipping (i.e. digitial distortion) and its not a good type of distortion. And since you normalize, then boost 6dB more, you are clipping at least some of your song.
What part of normalization don't you understand here? Normalizing finds the maximum amplitude in your song, subtracts that value from the Max peak possible (0dBfs), and adds that difference to your entire track. if your song already goes over 0, Normalizing does nothing.
And just because the master fader is set to "0" doesn't mean your song is set to 0dBfs!!... it just means the master fader is neither boosting or cutting your signal.
I understand what normalization does, and at first I wasn't doing what I described in my first post. But it sounds like I'm loosing a lot of "punch" when I'm normalizing.
I know that setting the master to 0 doesn't limit the sond to 0db, I was just explaining how I do it. I'm totally aware that I go over 0db, but what the use of the 6db if nobody is supposed to use them?
And I've gotta say that every actual commercial recording go over this limit.
I found a link to a wikipedia page when I was searching the forum:
http://en.wikipedia.org/wiki/Loudness_war
Well, that's really insteresting, but if I don't push the volume to the max, my songs sound even crappier
About compression, I'm sure it's the way to do it, but I'm not sure it's what I need. I like my song when I hear it in ableton live, and what I want is a wav file with my exact song, with the maximum volume (I know you're gonna say that normalizing does that, but I get something far lower than a commercial record).
EDIT: Oh, and another question, does a song clips when it reaches +6db or can it saturate before that limit?
-
solacerodgers
- Posts: 191
- Joined: Wed May 21, 2008 9:35 pm
- Location: myrtle beach
- Contact:
Actually i think you have some concepts misunderstood here so lets break em down for ya.
Going over 0db is not a bad thing, going over 0dbfs is because everything that goes over is gone. What does that mean. Well it affects transients most so that sharp snare attack can be said bye bye to and replaced with, well nothing. You are left with a digitaly clipped peice of audio that sounds a little worse. Now to the 6dbfs of headroom you were asking about the reason for this ( or any headroom ) is for room to grow when you get into the "mastering" stage. You can get your exact sound with little to no change in anything but volume without ever touching the normalize button. Its all about transparancy. Some compressors and limiters are made to introduce "tones" or character or earlier models and people like that because it adds warmth to digital audio but you can go right by that if you want and find some clean tone free tools. I am pretty sure the ones in ableton are made to sound like opto instead of vca or tube anyway so leave the -6dbfs to -4dbfs of head room and before you export or render it out just do this.
Download this file it contains a effects rack preset and a mastering limiter thats freeware. If the settings are not giving you the result you need you can turn the limiter up ( meaning the threshold down ) do not normalize on export. This is just basic stuff if you want real results pm me or see you local mastering engineer.
http://www.megaupload.com/?d=KZ3SBO8X
Going over 0db is not a bad thing, going over 0dbfs is because everything that goes over is gone. What does that mean. Well it affects transients most so that sharp snare attack can be said bye bye to and replaced with, well nothing. You are left with a digitaly clipped peice of audio that sounds a little worse. Now to the 6dbfs of headroom you were asking about the reason for this ( or any headroom ) is for room to grow when you get into the "mastering" stage. You can get your exact sound with little to no change in anything but volume without ever touching the normalize button. Its all about transparancy. Some compressors and limiters are made to introduce "tones" or character or earlier models and people like that because it adds warmth to digital audio but you can go right by that if you want and find some clean tone free tools. I am pretty sure the ones in ableton are made to sound like opto instead of vca or tube anyway so leave the -6dbfs to -4dbfs of head room and before you export or render it out just do this.
Download this file it contains a effects rack preset and a mastering limiter thats freeware. If the settings are not giving you the result you need you can turn the limiter up ( meaning the threshold down ) do not normalize on export. This is just basic stuff if you want real results pm me or see you local mastering engineer.
http://www.megaupload.com/?d=KZ3SBO8X
Well, I think I had understood that, except that I didn't know which was which between db and dbfs.
So, if I've got a track that never goes over 0db, I normalize it. I get a file with a peak at 0db/-6dbfs. I put this file in ableton live, and push the volume to, let's say 5,98db. That means that my render track will have a peak at 5,98db/-0,02dbfs, so it won't clip and I won't loose data, right?
I repeat, I don't want to do a real master on my tracks (just because I don't know anybody to do it).
Thanks for the links (laird and solacerodgers), I'm gonna test limiters and compressers, and one day, I promise, I will totally forget my method
. But what's so wrong about it if it doesn't clip?
So, if I've got a track that never goes over 0db, I normalize it. I get a file with a peak at 0db/-6dbfs. I put this file in ableton live, and push the volume to, let's say 5,98db. That means that my render track will have a peak at 5,98db/-0,02dbfs, so it won't clip and I won't loose data, right?
I repeat, I don't want to do a real master on my tracks (just because I don't know anybody to do it).
Thanks for the links (laird and solacerodgers), I'm gonna test limiters and compressers, and one day, I promise, I will totally forget my method
-
solacerodgers
- Posts: 191
- Joined: Wed May 21, 2008 9:35 pm
- Location: myrtle beach
- Contact:
where'd you get the 0dB = -6dBfs equation??
If you have an audio file, its measured in dBfs. 0 is the max.
If you have a sound in the air, its dBspl. as in the rock show was at 120 decibels
if you have audio encoded as an electric signal, its probably in dBu. You can go over 0 and get good distortion.
If you normalize, then boost your file by +6dBfs, you will get clipping. It might be unnoticeable to your ears... but something has been irretrieveably lost. Arguments can be made that a little clipping is really not so bad.
now, live is a little tricky with its 32bit floating point maths, but the bottom line is if you render a file thats going over 0, you have clipping.
Now, if you trust your ears, then fine, you can claim your way is better than good mixing and good use of compressors/limiters.
But in my opinion, its not good technique to clip your files so that you can compete in the loudness wars. There are better ways to compete.
If you have an audio file, its measured in dBfs. 0 is the max.
If you have a sound in the air, its dBspl. as in the rock show was at 120 decibels
if you have audio encoded as an electric signal, its probably in dBu. You can go over 0 and get good distortion.
If you normalize, then boost your file by +6dBfs, you will get clipping. It might be unnoticeable to your ears... but something has been irretrieveably lost. Arguments can be made that a little clipping is really not so bad.
now, live is a little tricky with its 32bit floating point maths, but the bottom line is if you render a file thats going over 0, you have clipping.
Now, if you trust your ears, then fine, you can claim your way is better than good mixing and good use of compressors/limiters.
But in my opinion, its not good technique to clip your files so that you can compete in the loudness wars. There are better ways to compete.
I'm talking about the live measuring system: it can encode a sound up to 6db, which is the maximum.laird wrote:where'd you get the 0dB = -6dBfs equation??
I'm not saying my way is better, I even said that I wish to get better ways, that's the point of my first question.laird wrote:now, live is a little tricky with its 32bit floating point maths, but the bottom line is if you render a file thats going over 0, you have clipping.
Now, if you trust your ears, then fine, you can claim your way is better than good mixing and good use of compressors/limiters.
But in my opinion, its not good technique to clip your files so that you can compete in the loudness wars. There are better ways to compete.
But it will be hard because I'm a total noob, and I need a lot of work to learn a way that will render something better only when I will have mastered it.
Moreover, can someone explain me why every commercial record goes over 0db? And if I understood laird well, all of theses records clip, even if they peak at 0,1db ? (I must be confused because solacerodgers said the opposite...)
So why everybody's doing it, and why does it sound well?
No, you've got this wrong. live cannot make a magical 16bit/44,1kHz .wav file that goes above 0dB. 0 is the max.MusicFlow wrote:I'm talking about the live measuring system: it can encode a sound up to 6db, which is the maximum.laird wrote:where'd you get the 0dB = -6dBfs equation??
Furthermore, commercial CDs do NOT GO ABOVE 0. They hit 0 for >3 samples in a row. Nowhere on a CD will you find a 17bit number (i.e. a number greater than 0dBfs)
Live's floating-point architecture makes things a little tricky.... you CAN go above 0dBfs INSIDE Live and not get clipping....
But the bottom line is if you go into the red on the master track, and then hit render, you are clipping.
Its not the volume that makes clipping bad, its the fact you are truncating what should be smooth waveforms and making them square waves.
Clipping your tube amplifier can also produce distortion, but this can be harmonic distortion. Digital clipping is not harmonic. it generally sounds bad, and thus should be avoided. not always, but in general.
See, in this bottom pic, you can imagine that the waveform should be some sorta sine wave, but not it looks more like a square wave. this is what you want to avoid. Going above 0dBfs when rendering will give you this:

Let me explain one thing a bit further:
Live shows you when the volume is going above 0dB. Everything turns red, right?
live is smarter than your average CD player. Inside live, you can go above 0dB. You cant do this outside Live.
Sorta like the "layers" inside of an Adobe Photoshop file. Once you print the photo on paper, there are no layers.
Once you print (render) to a 16 or 24bit file, that data that is going above 0dB is lost. And you can't get it back. its all 0s. And it'll probably sound bad. overall, the average volume will be louder, but at the expense of added distortion.
Compressors, tube amp silumaltors, limiters, all try do the same thing but without so much bad distortion.
Live shows you when the volume is going above 0dB. Everything turns red, right?
live is smarter than your average CD player. Inside live, you can go above 0dB. You cant do this outside Live.
Sorta like the "layers" inside of an Adobe Photoshop file. Once you print the photo on paper, there are no layers.
Once you print (render) to a 16 or 24bit file, that data that is going above 0dB is lost. And you can't get it back. its all 0s. And it'll probably sound bad. overall, the average volume will be louder, but at the expense of added distortion.
Compressors, tube amp silumaltors, limiters, all try do the same thing but without so much bad distortion.