| Author
|
headroom Odb?
|
ZilDoggo
Started Topics :
4
Posts :
663
Posted : Feb 28, 2005 14:44
|
ok.,
lemme help you guys out a bit.,
as far as i understand the cubase audio engine,
all the extra headroom is ABOVE the channels 0db point.
below this you have the normal 16 (or 24) bit accuracy. (no matter wether its float or not, the exact same bits are represented)
above 0db (on the channels) you get into the extra headroom bit.
then all channels are mixed in a 32 bit float space and arrive at the master outputs.
these also have a odb point but this one is the real one.
after this step the sound will be converted into the final bitdepth and any signal above 0db will be clipped.
so it is not very strange to have the kick play at or over 0db on the CHANNEL (altho it can make metering a bit hard).
you will in fact be avoiding certain types of errors due to loss of resolution by 'stretching' it into a higher bit depth.
but you should avoid clipping at the output stage as this is where the final signal is produced.
if the signal is too loud at the outputs, just lower them untill no clipping is visible.
also, dont drive the channels too hard as you may even clip the 32 bit space., (you need to make it REALY loud, but with a couple of plugins this should not be a problem)
as someone said, a low level at the output stage is not so bad if you export in 32 bit float.
if you export in 24 bits just make sure you are close to 0db
with 16 bits renders you should realy watch the levels and try to keep it as tight as possible cause every bit counts..,
and another thing, 32 bit float is NOT infinite.
pure garbage., there is no such thing as infinity in computers..,
anyone claiming the opposite is either a salesman or a noob,.
greets..,
aka., |
|
|
ZilDoggo
Started Topics :
4
Posts :
663
Posted : Feb 28, 2005 16:49
|
a little update,
i've done some checks in cubase and it seems that 32 bit float is A LOT OF resolution.
i made 2 small plugins in synthedit that boost, respectively cut level (quite extremely),.
i was able to reduce and then boost the level by more than 90db.
the resulting wave was identical to the original input wave.,
tried several other things, like boosting on the channel and reducing on the output buss.
same results, output wave is same as input wave.
in other words, dont even think about this kind of stuff., its good enough.,
thought you might want to know.,
greets.,
aka., |
|
|
|
fluct
Started Topics :
4
Posts :
1
Posted : Feb 28, 2005 17:31
|
|
thankyou all |
|
|
|
thockin
Started Topics :
1
Posts :
114
Posted : Feb 28, 2005 18:16
|
It's not infinite, but it's larger than you'll ever use.
Remember this: 1.0 is -0dB. The max number that a float can hold is 340282347000000000000000000000000000000.
|
|
|
|
br0d
IsraTrance Junior Member
Started Topics :
12
Posts :
355
Posted : Mar 1, 2005 15:30
|
Quote:
|
On 2005-02-27 13:41, Jupiter wrote:
Quote:
|
On 2005-02-27 13:27, br0d wrote:
at the outset, set the kick and bass to roughly -5dB and then adjust to taste from there.
at -12db you lose 1bit of resolution in a 16 bit recording. nothing huge to worry about.
i tend to use limiters on my master bus as a gauge during mixing, and then disable them for export. that way I can work to peak values of +3 to+6.
then just pull the master bus down a tad when exporting, to prevent clipping. sorted.
|
|
Again,you don't loose a bit at all.The bitdepht has nothing to do with the Volume in the mixer,it has to do with the Dynamic Range of the Audiosignal and that range will always be the same,no matter how loud you play that signal.Get it?
Nice greetings
Fred
|
|
Right, at -12dB before a conversion, I didn't say anything about setting the mixer to save bit depth. The "bit resolution warning" is for input sources coming through AD converters anyway, apps like SX work at 32bit internally.
And the -5 thing is an industry standard for setting starting levels, try it, it works. The goal is to REDUCE tech, not to increase it. So guidelines like that are useful.
And honestly I've only learned this stuff in passing from knowing some hardcore engineers, I've never bothered to sort out too many of the golden eared engineering details, because I find it too anal. A 15 bit signal sounds as good as a 16 bit signal, a 44.1kHz signal sounds as good as anything higher, and everything ever recorded on *any* DAW sounds better than the entire 80s punk era put together. Time spent worrying about "pristine elements" is creativity lost. Moreover, it shifts the focus from the right brain to your left brain, makes you less of an artist, and more of a science nerd.  |
|
|
|
br0d
IsraTrance Junior Member
Started Topics :
12
Posts :
355
Posted : Mar 1, 2005 15:39
|
Quote:
|
On 2005-02-28 14:44, ZilDoggo wrote:
ok.,
lemme help you guys out a bit.,
as far as i understand the cubase audio engine,
all the extra headroom is ABOVE the channels 0db point.
below this you have the normal 16 (or 24) bit accuracy. (no matter wether its float or not, the exact same bits are represented)
greets..,
aka.,
|
|
Yes in Cubase all of the headroom is above 0 dB on the instrument channels but not the master. Therefore, it is advised to work hot on the instrument channels, exceed 0dB on the master, and then pull down the master to avoid digital clipping. I don't personally do that because I find that strapping a limiter across the master and then watching its gain reduction tends to enforce a more even mixing practice, at least for me, ie, I granularize and tacke the problem at its source--the instrumental balances. Others may get more out of the "master yanking" method.
You will notice that many patches and VST instruments default to levels which seem ungodly loud, epsecially compared to older analog gear. This is why. it's really, really hard to clip an instrument channel in SX.
and 32 bit float has ungodly headroom, but it also takes up ungodly hard drive, twice that of 16 bit...i work at 24 bit as an acceptable compromise, as I am too lazy to move over to DVD-R for backups and archival.
HTH <--not trying to help, just sanitizing your pool |
|
|
|
|
