32 BIT FLOAT

a computer-related recording forum with user woes, how-to's and hints
Locked
rozzie
audio school graduate
Posts: 12
Joined: Fri Oct 01, 2004 3:08 am
Location: Ireland
Contact:

32 BIT FLOAT

Post by rozzie » Tue Dec 21, 2004 2:45 am

Hi,

I'm using Nuendo 2 with the Emu 1820m card.
I have the Emu set to record at 24bit, Now in the project setup window i can set the record format to 24bit or even 32bit float.
Should i use the 32bit float of just 24 bit when the card is set at 24 bit??
Thankz Roz

DoubleDownSound
gimme a little kick & snare
Posts: 89
Joined: Wed May 07, 2003 12:10 pm
Location: TEXAS
Contact:

Re: 32 BIT FLOAT

Post by DoubleDownSound » Tue Dec 21, 2004 7:48 am

I use Nuendo 1.5.1 with either the emu 1212m or 1820 (depending on which computer I am at) I only use the 32-bit float option with stereo tracks that are already mixed. Try it your way and report back, I'm curious
Don't judge Texas by one person who claims to be a Texan...

hammertime
re-cappin' neve
Posts: 722
Joined: Sat Nov 13, 2004 9:17 am

Re: 32 BIT FLOAT

Post by hammertime » Tue Dec 21, 2004 5:00 pm

I use it with cubase sx, and it sounds good to me, especially if you have some decent converters. Theoretically, at least, it shouldn't make too much difference, unless you do alot of processing on the file -- in which case, it should minimize quantization distortion from truncated word lengths. There was an article in Mix about a year ago, though, where the guy who founded Sonic Solutions argued that 24 bit fixed point sounded better than 32 bit float. I've recorded files back to back, and didn't notice any difference, but as I said the difference might become apparent if you repeatedly edit the file.

User avatar
Mr. Dipity
carpal tunnel
Posts: 1528
Joined: Mon Jun 16, 2003 11:29 am

Re: 32 BIT FLOAT

Post by Mr. Dipity » Tue Dec 21, 2004 9:07 pm

hammertime wrote:I use it with cubase sx, and it sounds good to me, especially if you have some decent converters. Theoretically, at least, it shouldn't make too much difference, unless you do alot of processing on the file -- in which case, it should minimize quantization distortion from truncated word lengths. There was an article in Mix about a year ago, though, where the guy who founded Sonic Solutions argued that 24 bit fixed point sounded better than 32 bit float. I've recorded files back to back, and didn't notice any difference, but as I said the difference might become apparent if you repeatedly edit the file.
Can anyone think of a decent process that just causes quantization distortion, and nothing else? No reduction in dynamic range, no other changes to the sound. It can be several steps, if you like.

brew
pushin' record
Posts: 284
Joined: Thu Aug 14, 2003 2:06 pm
Location: Brooklyn

Re: 32 BIT FLOAT

Post by brew » Wed Dec 22, 2004 8:50 am

Can anyone think of a decent process that just causes quantization distortion, and nothing else? No reduction in dynamic range, no other changes to the sound. It can be several steps, if you like.
A decent process? Not sure what that means, but reducing the wordlength from 24 bit to 16 bit without dither will cause quantization distortion as will using A/D's without dither (don't think these exist anymore). You can't do anything without changing the sound because quantization distortion changes the sound![/quote]

User avatar
Mr. Dipity
carpal tunnel
Posts: 1528
Joined: Mon Jun 16, 2003 11:29 am

Re: 32 BIT FLOAT

Post by Mr. Dipity » Wed Dec 22, 2004 9:21 am

brew wrote:
Can anyone think of a decent process that just causes quantization distortion, and nothing else? No reduction in dynamic range, no other changes to the sound. It can be several steps, if you like.
A decent process? Not sure what that means, but reducing the wordlength from 24 bit to 16 bit without dither will cause quantization distortion as will using A/D's without dither (don't think these exist anymore). You can't do anything without changing the sound because quantization distortion changes the sound!
[/quote]

Reducing the dynamic range also reduces the dynamic range, however.

i guess what I'm looking for is a process that I could build in Reaktor or set up with creative use of existing plugins, so I could set up some kind of batch process, repeat it a few tens of times and answer this question once and for all.

User avatar
syrupcore
deaf.
Posts: 1793
Joined: Mon Mar 08, 2004 4:40 am
Location: Portland, Oregon
Contact:

Re: 32 BIT FLOAT

Post by syrupcore » Thu Jan 27, 2005 8:58 pm

sserendipity wrote:i guess what I'm looking for is a process that I could build in Reaktor or set up with creative use of existing plugins, so I could set up some kind of batch process, repeat it a few tens of times and answer this question once and for all.
sser, did you ever come up with a process? really good idea.

User avatar
andyg666
re-cappin' neve
Posts: 669
Joined: Tue Jun 17, 2003 9:25 am
Location: Witchtown, MA
Contact:

Re: 32 BIT FLOAT

Post by andyg666 » Fri Jan 28, 2005 6:00 am

i'd use 24 bit rather than 32 bit float. a lot of programs don't know how to interpret 32 bit float data. for instance, if i record something in cool edit pro and save it as a 32 bit float file, then try to open it in recycle, it doesn't work. i end up having to re-save it as a 24 bit file. i don't think the difference in sound quality is going to be noticeable unless you have a PERFECT room and PERFECT ears. just go with the format that's more universally accepted. IMHO.

Locked

Who is online

Users browsing this forum: No registered users and 6 guests