Normalize? Mastering? Need help.

actually it wouldnt take much at all to technically clip 32bit floating point. just two already at close to 0dbfs, regardless of dynamics, would do it. to get that you just need to turn up the faders on the tracks concerned, its not related to dynamic range at all.

all thats somewhat redundant as 32bit floating point by definition cannot clip. the floating point allows values over 0dbfs to be measured. the problem occurs when you play it out as at some point your DA converter will probably need to stick it to fixed point to convert and at this stage anything above 0dbfs will be lost and represented as flat lines....and awful sound.

to ensure your converters are ok regardless of rates, try to ensure peak on the main out isnt above -0.5dbfs. poor converters can clip below 0dbfs and this should protect this just incase.
 
Word, yo. :bigeyes:

Floating point is another "breakthrough" that was put in to be able to record *QUIETER* with high resolution - Not louder. It has an almost immeasurable noise floor and DOWNWARD dynamic range.
 
neilwight said:
actually it wouldnt take much at all to technically clip 32bit floating point. just two already at close to 0dbfs, regardless of dynamics, would do it. to get that you just need to turn up the faders on the tracks concerned, its not related to dynamic range at all.
Click to expand...



It might come off as confusing when you state that it can clip witin 32-bit floats using two track, and then go on to say it can't clip.

Like I said : Internally it won't clip.
The master out is a different story cause that wil clip
The key word here is floating point
Try getting it to clip internallly.
a quick test in Nuendo.
-29 identical tracks all heavilly limited and peaking at "0dB"
-smacked ALL the channel fader to the top +6.02 dB
- Pulled down the masterfader til it was peaking at -1dB
-Exported the output and brought it back into the project.
-Play it back next to one of the original track.
-Adjust the levels.Phase invert it and you get total phase-cancellation

What does that tell you about the files Any internal clipping?
In other words massive headroom

I already stated that the master out will clip as the data has the left the 32-bit float environment and is converted to INTEGER.
Fixed point is something else.

all thats somewhat redundant as 32bit floating point by definition cannot clip. the floating point allows values over 0dbfs to be measured. the problem occurs when you play it out as at some point your DA converter will probably need to stick it to fixed point to convert and at this stage anything above 0dbfs will be lost and represented as flat lines....and awful sound.
Click to expand...

There is no value over 0dBFS in digital
The reason 24-bit fullscale won't overflow in a 32-bit float environment is simple enough: there's plenty of room for piling.

The reason it cannot clip has nothing to do with the abililty to "measure" values above 0dBFS really.

The way 32-bit floating point is treated it's always a tradeoff between size and precision
It has everything to do with the varibles ability to store large numbers and small numbers. Size and precision.

It can scale the value without sacrificing resolution.

I'm not sure what you point is., as I already state that the only place it will clip is the master out.
Yet you start yor post with saying you can clip within 32-bit floating point environment with only 2 track.
Or are you talking about the master-out?

If you think that clipping distortion is awfull, you should hear the result of overflows when it's not handled and folds.
:)
 
MASSIVE Mastering said:
Word, yo. :bigeyes:

Floating point is another "breakthrough" that was put in to be able to record *QUIETER* with high resolution - Not louder. It has an almost immeasurable noise floor and DOWNWARD dynamic range.
Click to expand...

To be precise, this(32-bit float) has nothing to do with the actual recording or the ability to record louder or quieter. It's about the processing.

Using 32-bit float processing has nothing whatsoever to do with you sample bit depth.
 
edited,
.
this is not an attempt to sidestep but purely signifies, sadly, a lack of faith at this end that anything will be heard.
 
Last edited by a moderator:
MASSIVE Mastering said:
Normalizing simply adjusts the peak of a file to match a specific level.
Click to expand...

can u tell me an example or something, so i could visualize/understand it better?

i would really like to understand the concept of normalize...cuz i also thought that it just evens out all the files to have the same volume...but i'm hearing that that's not the case at all, so i'd really like to know :-)


_chris
 
It has nothing to do with apparent volume.

If you have a track with a 60dB dynamic range (noise floor to peak) that peaks at -10dBFS, the noise floor is at -70dBFS.

If you normalize that file to -1dBFS, the noise floor will be at -61dBFS.

It's a peak volume control - Nothing more.

It WILL make every file have the same PEAK. Again, that has nothing at all to do with having the same LOUDNESS.

Every mix is going to have a different crest factor - Normalizing doesn't take that into account.

I shouldn't say that - There are a few programs that do... Very badly, I might add. RMS normalization algorithms are one of the biggest wastes of processing power I've ever heard. They're just taking a "guess" at it - You're better off just trusting your ears.
 
Last edited by a moderator:
MASSIVE Mastering said:
If you have a track with a 60dB dynamic range (noise floor to peak) that peaks at -10dBFS, the noise floor is at -70dBFS...
Click to expand...

how would u know if the track is one with "60dB dynamic range"?

would there be a case where it's higher/lower than 60dB?
how would that come about?

sorry if it's a dumb question(s)...but i like knowing all the angles of things.
lol

i'm kinda anal like that...heh

i just like making sure i understand everything...ya know?


_chris
 
LateArrival101 said:


how would u know if the track is one with "60dB dynamic range"?
Click to expand...


that is not the point... it doesn't matter... it was just an illustrative example. The numbers were there to explain the concept.
 
i know...i know...

but it just didn't really hit me cuz i'm thinking like this:

if the peak IS -10db....how would you automatically know that it's a 60db range?

how would you really know what the floor is? couldn't it be LESS than
-70...or vice-versa?

sorry...it's just really confusin' me.
i know the concept of numbers and stuff.
i'm actually very GOOD at math.
lololol

but yeah....i'm just still confused i guess.


_chris
 
if the peak IS -10db....how would you automatically know that it's a 60db range?
Click to expand...
You'd know if the noise floor was at -70dBFS with a -10dBFS peak.

Is that what you're looking for?

It could be more, it could be less - A noisy room, noisy gear, noisy preamps, high noise floor. Maybe 30dB. Analog tape, 60-70dB.

24-bit digital... 144dB

And yet, everyone wants to record loud for some reason... But that's for another thread...
 
Last edited by a moderator:
You must log in or register to reply here.
Back
Top