| Author
|
Avoiding clipping on the master
|
Fingax
Cosmic Station
Started Topics :
82
Posts :
1235
Posted : Apr 17, 2006 18:37
|
 bit revolution
|
|
|
e-motion
IsraTrance Full Member
Started Topics :
71
Posts :
933
Posted : Apr 17, 2006 18:39
|
just played it in my non 32bit soundcard. sounds funny... altough it's better than sub6 music
*off topic*  |
|
|
|
UnderTow
Started Topics :
9
Posts :
1448
Posted : Apr 17, 2006 19:51
|
I'm going to have to disagree with Colin here.
Although he is right about floating point audio in theory , not all plugins use floating point maths and some can't handle signals above 0dB FS properly. Also, alot of plugins are not linear in their response: They sound better when you give them headroom to operate with.
Also, although you are not "using all the bits" when your signal isn't peaking at 0 db FS, the noise floor is at arround -144 dB FS (at 24 bit). Negligable. On the other hand, with every gain change, you are re-quantising the signal thus adding more quantization distortion. It isn't much but why add it when you can avoid it?
What does this all mean in practise? For me it means that I stick to old school mixing practises. I keep all my signals way below 0 dB FS. That means all channels, all busses and the main stereo out before the "mastering plugins".
The "mastering plugins" can make up any lack of level easily and are designed to do gain changes with the best precision available. (64 bit for the good ones).
So if the stereo out (with the FX bin disabled) goes above (or even close to) 0 dB FS, I fix the mix. I don't touch the master fader.
Another advantage of working this way is that if you ever work on a nice big analogue/digital console (Or ProTools). you don't have to learn any new tricks!
UnderTow
|
|
|
|
Colin OOOD
Moderator
Started Topics :
95
Posts :
5380
Posted : Apr 17, 2006 20:00
|
|
fregle
IsraTrance Junior Member
Started Topics :
11
Posts :
982
Posted : Apr 17, 2006 21:53
|
|
ok, so i got the 32-bit stuff, but i have one more question, normalising is a big nono in all the tutorials i have read, but they all talk about normalising when the level is below the maximum allowed by the headroom. Now u are talking about normalising a 32-bit file that goes beyond the headroom to fit the headroom provided by 16-bit files, right? Is normalising in that situation (lets call it normalising 'down') as destructive as normalising when the level is below the allowed headroom (or, normalising 'up')? or is it in fact better than dithering? |
|
|
|
Kitnam
Mantik
Started Topics :
110
Posts :
1151
Posted : Apr 17, 2006 23:41
|
thx colin  |
|
|
|
Meta
Meta/Boomslang
Started Topics :
24
Posts :
1045
Posted : Apr 17, 2006 23:55
|
As far as mixing and avoiding clipping goes, I usually create a few group channels to route everything into, so at the end stage I can just lower 3 or 4 faders at proportionate levels instead of lowering huge ranges.
So, I usually have Kick & Bass in one group, drums in another group, synths in third group, if neccessary, a fourth one with FX/samples/whatever.
I connect the sendFX bus levels to the synth group channel level to keep everything proportionate.
This has worked pretty well for me, as far as getting stuff as hot as possible without clipping, and also for narrowing where certain peaks or whatnot are coming from.
 http://soundcloud.com/aeon604
http://www.metaekstasis.com/
http://the1134.com/ |
|
|
|
Zeyon
Started Topics :
1
Posts :
145
Posted : Apr 18, 2006 00:49
|
Hm.. I haven't quite understood the thing about the +12dB wav you made, Colin, although I certainly hear the point (It's pretty astonishing, normally when you have stuff that's clipping you can't do anything about it because it will always have the distortion), so, can you (Or one af the others) tell me again what the 32bit float does to the audio?
Sorry for me being such a n00b, but I'm not really into such technical stuff (Yet)..
//Zeyon. |
|
|
|
UnderTow
Started Topics :
9
Posts :
1448
Posted : Apr 18, 2006 01:48
|
Zeyon, 24 or 16 bit audio uses fixed bits. When all the bits are on (1's), you are at 0 dB Full Scale (FS). When all the bits are off (0's) you are at -96 dB FS and -144 dB FS respectively.
Anything above 0 db FS clips and sounds bad.
In 32 or 64 bit Floating Point (FP) Audio, there are 24 and 52 bits respectively used for the resolution of the audio (the Mantissa) and 8 or 12 bit respectively used for scaling the level of the signal up and down (the exponent). (This is slightly simplified).
So when you go over 0 dB FS in FP audio, the signal is still encoded with 24 or 52 bits but the exponent scales the level up and down without the signal clipping.
Does that clear things up?
UnderTow |
|
|
|
piko_bianko
Oxya
Started Topics :
57
Posts :
974
Posted : Apr 18, 2006 02:38
|
a lil bit irrelevant i suppose, BUT, i've recently heard from some Italian and Spanish pioneers of psy (not actually speaking directly to me) that final export should be near -12 db having kick's peak around there too, and everything else mixed due to this volume. they say low db leave more space for mids and not having a a great mid fuss as like mixing near -0.2 -0.1 db... aaand after that you can higher your track's db to max
additionally to that it was said that final volume should be -6db so it can be easily mastered in order to be released...
i'm not asking artists only (who might already know) but everyone can share his knowledge aside or against that
thank you, piko.
extreme |
|
|
|
Spindrift
Spindrift
Started Topics :
33
Posts :
1560
Posted : Apr 18, 2006 04:11
|
Quote:
|
On 2006-04-18 02:38, piko_bianko wrote:
a lil bit irrelevant i suppose, BUT, i've recently heard from some Italian and Spanish pioneers of psy (not actually speaking directly to me) that final export should be near -12 db having kick's peak around there too, and everything else mixed due to this volume. they say low db leave more space for mids and not having a a great mid fuss as like mixing near -0.2 -0.1 db... aaand after that you can higher your track's db to max
additionally to that it was said that final volume should be -6db so it can be easily mastered in order to be released...
i'm not asking artists only (who might already know) but everyone can share his knowledge aside or against that
thank you, piko.
|
|
The standard when it comes to digital ausio is to say that -12 db FS should be considered to be unity gain or 0 dBu.
That leaves you with a amount of headroom about the same as on an analouge desk on a 16 bit system like a DAT.
But with 32 bit float that would't really be relevant in theory as colin said.
And for trance music I never needed 12 db when I used DAT for recording the master.
The dynamics of electronic music is very controllable so 6-8 db should usually be plenty enough.
That you should get clearer mids from leaving more headroom to then bring it up again is not something I experienced and I find it highly unlikely.
The way I do it is pull my faders in logic down to about -6 db which usually is about right to make a binch of tracks with kick and bass aend up around 0 db.
Then after playing thru the track before recording the master will show how much over/under the highest peak was.
I then do a select all and pull the faders up or down by the amount the master is off so I leave about -0.3 db.
(``·.¸(``·.¸(``·.¸¸.·`´)¸.·`´)¸.·`´)
« .....www.ResonantEarth.com..... »
(¸.·`´(¸.·`´(¸.·`´``·.¸)``·.¸)``·.¸)
http://www.myspace.com/spindriftsounds
http://www.myspace.com/resonantearth |
|
|
|
Zeyon
Started Topics :
1
Posts :
145
Posted : Apr 18, 2006 13:37
|
|
Oh, yes, now I understand it, thanks alot..
//Zeyon. |
|
|
|
sy000321
IsraTrance Junior Member
Started Topics :
46
Posts :
1142
Posted : Apr 18, 2006 14:30
|
why instead of mixing everything at 0dbs and then setting the fader down -6dbs, why not start with the bass/kick peaking at -6dbs (that's how i do it)?
roll a joint or STFU :) |
|
|
|
Colin OOOD
Moderator
Started Topics :
95
Posts :
5380
Posted : Apr 18, 2006 17:31
|
|
