Avid Pro Audio Community

Avid Pro Audio Community (https://duc.avid.com/index.php)
-   macOS (https://duc.avid.com/forumdisplay.php?f=92)
-   -   24 or 32 bit float (https://duc.avid.com/showthread.php?t=316309)

Bender 01-26-2012 03:56 PM

24 or 32 bit float
 
I have always used 24 bit
Now I have the 32 bit floating point option.
I always checked stereo interleaved when I bounced and now it is an option at session setup as well.
What are the benefits of these options

Thanks

Craig F 01-26-2012 04:04 PM

Re: 24 0r 32 bit float
 
tracking, none the converters are still 24 bit
mix, there is some benefit now can you hear that on your system in your room, I don't know

stevecaboose 02-02-2012 01:13 AM

Re: 24 0r 32 bit float
 
Lets be real here. Like Craig said (and hes right) your converters are only going to convert at 16 or 24 bit, not 32. I personally think that making a Pro Tools session at 32 is pointless. In the end your going to be at 24 or 16 bits anyway. And you are not going to be able to hear a quality difference between 24bit and 32bit. The human ear cannot hear a difference what so ever after 20bits. The only thing you are doing by making a session at 32bits is using a lot more disk space and also demanding more processing power from your computer.

Franco1967 02-02-2012 02:11 AM

Re: 24 0r 32 bit float
 
What you are not seeing is the major fact, that the plugins are using 32-Bit and you have an enhanced headroom. By the way, the point is not, if anyone can hear 32-Bit, but the quantization noise is lower when you are working with lower levels. Thatīs a thing everyone can hear. The SNR is more enhanced than with lower Bit-Levels.

janmuths 02-02-2012 06:44 PM

Re: 24 0r 32 bit float
 
Quote:

the quantization noise is lower when you are working with lower levels. Thatīs a thing everyone can hear. The SNR is more enhanced than with lower Bit-Levels.
Your audio still goes in and out thru 24 bit converters. Thus giving you a fix quantisation error before the signal even reaches PT. And even 24-bit converters have a quantisation error way below the noise floor of all your analogue front end: mics, pres, amps, speakers (assuming proper gain structure). Recording audio in 32 bit makes little sense to me.

You do have a valid point about processing. Higher bit-rates make perfect sense when summing or processing. Plug-ins should be of higher bit-rate and they are: 48-bit linear in TDM, 32-bit float in most native DAWs. But recording audio files to disk... I really don't see the point.

Quote:

Thatīs a thing everyone can hear.
Are you sure you can hear the quantisation-error level difference between two audio files recorded thru the same 24-bit converter, one recorded in 32-bit float, the other recorded in 24-bit linear?
I'd be seriously impressed if you identified the 32-bit-float file in a blind test.

My two cents.

Jan

Bob Olhsson 02-03-2012 10:11 AM

Re: 24 0r 32 bit float
 
My 16 bit tests showed that files first dithered to 24 bits sounded better than 32 bit float files when they were were subsequently dithered to 16. Writing processing to 32 float makes sense if you are going to be doing radical additional processing or gain changes but dithered 24 seems to retain image size and low level harmonics better if you are only making minor changes.

64 bit float or 32 bit fixed intermediate process files are a different story but 32 float is still for all intensive purposes a truncated 24 bit file.

So Serious 02-03-2012 12:14 PM

Re: 24 0r 32 bit float
 
Quote:

Originally Posted by Bob Olhsson (Post 1897531)
My 16 bit tests showed that files first dithered to 24 bits sounded better than 32 bit float files when they were were subsequently dithered to 16. Writing processing to 32 float makes sense if you are going to be doing radical additional processing or gain changes but dithered 24 seems to retain image size and low level harmonics better if you are only making minor changes.

64 bit float or 32 bit fixed intermediate process files are a different story but 32 float is still for all intensive purposes a truncated 24 bit file.


Bob if you dont mind and just to be clear and dumb it down a little , as of protools 10 do you set your session up as 24bit or 32 float?, interleaved or no?
if 32 , do you dither to 24bit first/, then to 16 bit when you make a cd? Whats your process to dither from 32 to 24, is it with the plugin or another method?

Drew Mazurek 02-03-2012 12:24 PM

Re: 24 0r 32 bit float
 
I'm not Bob but at this point there is NO reason to start off a session at 32 bit and record into that way. Reason being there are no 32 bit converters.

The only reason to change to 32 bit is if you're going to be doing MAJOR clip gain and/or Audiosuite processes at Bob mentioned where the extra headroom comes in handy.

So Serious 02-03-2012 01:33 PM

Re: 24 0r 32 bit float
 
Quote:

Originally Posted by dmazurek (Post 1897586)
I'm not Bob but at this point there is NO reason to start off a session at 32 bit and record into that way. Reason being there are no 32 bit converters.

The only reason to change to 32 bit is if you're going to be doing MAJOR clip gain and/or Audiosuite processes at Bob mentioned where the extra headroom comes in handy.

So then it is probably a good idea that if i wanted to i could track in a TDM system, I could then mix the same session in native-32bit float to get the benefits that way? Of course in native I would be using a third party converter.

daeron80 02-03-2012 02:19 PM

Re: 24 or 32 bit float
 
Before you can even understand the answer to that question, you need to understand the basic differences between fixed-point and floating-point calculations and what advantages and disadvantages each has for digital audio. They're different animals and can't be compared head-to-head. Native versions of Pro Tools are going to do all their processing at 32-float regardless of your session settings, anyway. It's only about how the files on the disk will be treated.

Bob O is right (you can pretty much count on that around here in general :-)) that 32-bit float offers no greater resolution overall compared to 24-bit fixed. It offers a kind of variable or flexible resolution over a wider dynamic range at the expense of - I don't know exactly how best to express it - consistent precision? Something like that. Its main advantage is that it doesn't need to be dithered after each digital process. That means, if you do lots of Audio Suite processing or printing effects, layer after layer of multiple processes on the same files, then more detail may be preserved if the session is set to 32-float. Maybe. Remember that the session is going to take up 1/3 more space, and you will be able to play 1/3 fewer tracks per disk. So it darn well better sound enough better to make it worth that trade.

I can see 32-float sessions being an advantage in film work and sound design in some cases. That's about it.

A good book explaining all that in more detail is "Digital Audio Explained: For The Audio Engineer" by Nika Aldrich. Kind of expensive book but totally worth it. Your workflow will thank you henceforth and evermore.


All times are GMT -7. The time now is 01:21 PM.

Powered by: vBulletin, Copyright ©2000 - 2008, Jelsoft Enterprises Limited. Forum Hosted By: URLJet.com