Avid Pro Audio Community

Avid Pro Audio Community

How to Join & Post  •  Community Terms of Use  •  Help Us Help You

Knowledge Base Search  •  Community Search  •  Learn & Support


Avid Home Page

Go Back   Avid Pro Audio Community > Pro Tools Software > macOS

Reply
 
Thread Tools Search this Thread Display Modes
  #21  
Old 01-14-2013, 12:00 AM
zolko zolko is offline
Member
 
Join Date: Jan 2013
Location: Poland
Posts: 107
Default Re: 32 Bit Float

Oh please. Never heard of digital zoom? Feature in cameras that makes 16Mpx out of 4Mpx crop?
It was annoying when given (badly recorded anyway) 32bit float files for mix/master to PT7/8 I had to use another program to convert them. This is one of the reasons. If somebody implement 64 float and 384kHz files, the rest would do it either. This is then what thay nad then we call Progress and Hyper Super Duper High Definition :)
Reply With Quote
  #22  
Old 01-14-2013, 03:47 AM
Greg Malcangi Greg Malcangi is offline
Member
 
Join Date: Dec 1999
Location: UK
Posts: 491
Default Re: 32 Bit Float

Quote:
Originally Posted by StewartFang View Post
Oh please. Of course they sound different, just like 24 bit and 16 bit sound different.

DO you really think that the option was made available soley for those that do not know how to gain stage...REALLY?
There are definite advantages to higher bit depths for signal paths and processing. However, for you to say you hear a difference between 24bit and 32bit float when recording means either 1. You have a very poorly designed or defective ADC or 2. You are fooling yourself or 3. You are trolling. Unless you are trolling, maybe learning the basics of digital audio theory will help you to figure out which of the other two options is causing the difference you are hearing.

G
Reply With Quote
  #23  
Old 01-14-2013, 06:26 AM
CME CME is offline
Member
 
Join Date: Sep 2005
Location: Ardmore OK
Posts: 3,205
Default 32 Bit Float

Yep. Unless you are using an interface with 32-bit converters you're not gaining anything by recording at 32-bit, but losing disc space. Albeit with drives these days it's not much of a concern. Unless maybe like me you record hour and a half long sessions with 20+ tracks.

I will say you're right about 16 and 24 bit though. Most converters can be ran in either mode and that will actually make a difference.
__________________
2009 Mac Pro - Dual X5650 Xeon 6-core 2.66Ghz procs, 48GB of Ram, SATA 3 and USB 3.0 PCIe cards. R9 280X video card. macOS Sierra. Harrison Mixbus 32C, Studio One 3, Reaper, and pro tools 12 when I have to.
Reply With Quote
  #24  
Old 01-14-2013, 12:50 PM
Greg Malcangi Greg Malcangi is offline
Member
 
Join Date: Dec 1999
Location: UK
Posts: 491
Default Re: 32 Bit Float

Quote:
Originally Posted by StewartFang View Post
You guys are hilarious...
If you think you can hear the difference between 16bits of signal in a 16bit file format and the same 16bits in a 24bit or 32bit format file then you are the hilarious one. Good luck with that!

Quote:
Originally Posted by CME View Post
Yep. Unless you are using an interface with 32-bit converters you're not gaining anything by recording at 32-bit, but losing disc space.
There's really no such thing as a 24bit converter, let alone a 32bit one. They might write 24 or 32bit files but none of them can actually resolve 24bits (or 32bits) of audio data. The best ones are around 20bits. So, with a 24bit converter the least significant 4 or so bits are just noise. Of course, in a real world recording situation, the last 6-16 bits are nothing more than noise floor and even with a finished mix the last 14+ bits of data are just noise.

G
Reply With Quote
  #25  
Old 01-14-2013, 02:54 PM
dr_daw's Avatar
dr_daw dr_daw is offline
Member
 
Join Date: Oct 2012
Location: Cariboo, BC
Posts: 1,215
Default Re: 32 Bit Float

Quote:
Originally Posted by feliponk View Post
Guys, don't forget about this link above that 'feliponk' gave us. It show's how 32bit float works, I think that's what's being lost here.

"There's really no such thing as a 24bit converter, let alone a 32bit one. They might write 24 or 32bit files but none of them can actually resolve 24bits (or 32bits) of audio data. The best ones are around 20bits."

The above statement is untrue, there are MOST definitely 24bit ADC's, texas instruments makes skads(http://www.analog.com/en/analog-to-d...s/product.html). As for whether or not they are used in YOUR ADC, I don't know that. Regardless, the fundamentals of digital conversion and processing are getting lost here.

1 bit = 6db

So bit rate is directly proportional to headroom, the more headroom means the more signal available for sampling. The whole premis behind higher bit rates was to give us the ability to capture MORE of the the signal with a 6db buffer zone. So, if you're tracking with peaks hitting -6dbfs @ 16bits, you're really only getting 15bits, so that meant when we transfer this to CD it's not "true" 16bit that last bit is nothing(0).

With that in mind, you record @ 24bits with the same -6db rule, you've now effectively captured 23bits of audio signal.

I don't use 32bit because I believe in proper gain staging, and know that 32bit is just a conversion in the box and not a true bit depth(so I've been skeptical). But, I am going to try it to hear if it makes a difference. Problem being, that if you know how to gain stage, and move from 24bit to 32bit; it will most definintely sound better. You would have WAY more headroom to push your internal processing, then when it's dithered back to 24bit for mastering there will be LESS error bits, and even less when it's finally dithered down to 16bits. In the end, math and physics tells me it's going to be a better sounding file.

What's funny in this conversation is that no one has mentioned sample rates, that's where the true picture is captured. Recording at 24bit @ 44.1kHz is going to sound WAY different than 24bit @ 96kHz, you may not 'hear' the difference, but you will feel it...in your ears
__________________
FBR recording - https://www.facebook.com/fromthebase...cordsrecording
@FBRrecording - Instagram
Rig
MBP 15" High Sierra 10.13.3, i7 2.6Ghz, 8Gb DDR, 750Gb 7200rpm, PT 2018.10, Focusrite Saffire Pro 40, Octopre MkII, Glyph GPT50, Avalon 737sp, AMS Neve 1073's, API 512c, DW drums, Fender Guitars/Amps, AKG, Rode, Sennheiser, Neumann U87 & KM184's, Coles 4038, Equator D5 monitors
Reply With Quote
  #26  
Old 01-14-2013, 03:27 PM
Top Jimmy's Avatar
Top Jimmy Top Jimmy is offline
Moderator
 
Join Date: Oct 2005
Location: Spokane, WA
Posts: 5,028
Default 32 Bit Float

Quote:
Originally Posted by StewartFang View Post
You guys are hilarious...

"there's advantages...but they aren't sonic"

"I agree that 16 sounds different than 24...but there's noooowaaay that 32 sounds different than 24"

"32 is just a novelty for prosumers"



are you guys REALLY gonna stick with that?


A pro application making 32 bit recording just to appease prosumers?

YES you can hear a difference when recording 32 bit files. WHy? Who cares.
Stewie, do you even know that 32-bit floating point only quantizes 24 bits worth of signal? The other 8 bits are just for the exponent. The have absolutely no signal correlation beyond establishing the peak sample value.

The only true difference between 24 bit fixed and 32 bit float is a fixed or variable quantization noise floor.


Buttdialed by me & Tapatalk
__________________
James Cadwallader

Hackintosh - Gigabyte Z77X-UP5TH, Intel Core i7-3770K, 32GB 1600Mhz DDR3, 2x SATA 6Gb/s 240GB SSD, 3x SATA 6Gb/s 1TB HDD, 1x Glyph 2TB USB3 HDD, Sierra 10.12.6.

Pro Tools HD 2018.12, Focusrite Red 8Pre, Mbox 3 (Mac)
Pro Tools 11.3.2, Mbox2 (Win 7)

Presonus Faderport, Pro Tools | Control
Reply With Quote
  #27  
Old 01-14-2013, 04:32 PM
Greg Malcangi Greg Malcangi is offline
Member
 
Join Date: Dec 1999
Location: UK
Posts: 491
Default Re: 32 Bit Float

Quote:
Originally Posted by dr_daw View Post
"There's really no such thing as a 24bit converter, let alone a 32bit one. They might write 24 or 32bit files but none of them can actually resolve 24bits (or 32bits) of audio data. The best ones are around 20bits."

The above statement is untrue, there are MOST definitely 24bit ADC's, texas instruments makes skads(http://www.analog.com/en/analog-to-d...s/product.html). As for whether or not they are used in YOUR ADC, I don't know that. Regardless, the fundamentals of digital conversion and processing are getting lost here.

1 bit = 6db
Oh dear, have you even read what you have linked to?

"The AD7760 is a high performance, 24-bit Σ-Δ analog-to-digital converter (ADC). It combines wide input bandwidth and high speed with the benefits of Σ-Δ conversion to achieve a performance of 100 dB SNR at 2.5 MSPS"

So with a SNR of 100dB, from your own quote 1bit = 6dB, 100 divided by 6(dB) = 16.7bits. By it's own specifications, this chip actually resolves less than 17bits of audio signal, the remaining 7 bits in the 24bit file it outputs contains nothing other than it's own internal noise floor! In other words, this is a 17bit converter chip outputting a 24bit format file.

Quote:
Originally Posted by dr_daw View Post
What's funny in this conversation is that no one has mentioned sample rates, that's where the true picture is captured. Recording at 24bit @ 44.1kHz is going to sound WAY different than 24bit @ 96kHz, you may not 'hear' the difference, but you will feel it...in your ears
Really, who told you that? I hope you didn't pay for that useless piece of mis-information? The truth is of course that providing there are more than 2 samples taken of a wave cycle it can be reproduced perfectly. Sampling a wave cycle 10 or 5000 times cannot give you a more "true picture" than sampling it 3 times because 3 samples already gives you the truest picture possible! This really is basic stuff! Have you not even heard of the Nyquist-Shannon Sampling Theorem?

If you're going to call yourself or act as a recording engineer, mix engineer or whatever, you really should have at least a little fundamental knowledge of the engineering part of the job description!

G
Reply With Quote
  #28  
Old 01-14-2013, 04:35 PM
CME CME is offline
Member
 
Join Date: Sep 2005
Location: Ardmore OK
Posts: 3,205
Default 32 Bit Float

As been pointed out. Most interfaces these days use 24-bit converters. And due to the noise floor and limitations in analog stages we aren't even using the full 24-bits. No matter what, you're not going to get any more data than 24-bits out of a 24-bit converter. Period.

Most converters can also do 16-bit. But they switch into 16-bit mode. It's not just pro tools recording less bits.

Once you done recording and are processing the file, 32-bit can have advantages. And sample rate does have much more impact on sound quality, but that can be negligible and is probably due more to the performance of the converters than the actual recorded file.
__________________
2009 Mac Pro - Dual X5650 Xeon 6-core 2.66Ghz procs, 48GB of Ram, SATA 3 and USB 3.0 PCIe cards. R9 280X video card. macOS Sierra. Harrison Mixbus 32C, Studio One 3, Reaper, and pro tools 12 when I have to.
Reply With Quote
  #29  
Old 01-14-2013, 07:13 PM
dr_daw's Avatar
dr_daw dr_daw is offline
Member
 
Join Date: Oct 2012
Location: Cariboo, BC
Posts: 1,215
Default Re: 32 Bit Float

Greg,

I'm not going to resort to mud slinging, you're certainly right about that chip. For that I appoligize and shouldn't have used that as an example. I appreciate your help in pointing out my errors. I will be more deligeant in the future.

As for sample rate, I'd love some more insight on that one. Nyquist relates to the fact that the signal must be sampled a min of twice the highest frequency. That is only if you want to accurately reproduce the signal with MIN aliasing. There are benefits to oversampling. As we all know there are harmonics that can get missed during that principle of only 2xfrequency. What bit rate and sample rate do you use? What device?

My comment about 'feeling' the difference was a joke...just to be clear. I'm going to keep recording at 24bit 48kHz because that's what works for me. I'm also going to keep calling myself an engineer as well...probably more now :P (that is also a joke...I can really only professionally call myself a Linemen :P)
Reply With Quote
  #30  
Old 01-15-2013, 01:36 AM
Greg Malcangi Greg Malcangi is offline
Member
 
Join Date: Dec 1999
Location: UK
Posts: 491
Default Re: 32 Bit Float

Quote:
Originally Posted by dr_daw View Post
I'm not going to resort to mud slinging, you're certainly right about that chip. For that I appoligize and shouldn't have used that as an example.
You already resorted to mud slinging when you said my statement was "untrue". I appreciate and accept your apology though. Some of digital audio theory can be quite counter-intuitive and I've certainly misinterpreted parts of it myself in the past. The simple fact is that you won't find an example of an ADC which resolves more than about 20bits of data. The theoretical maximum is about 21bits, beyond that you would need to change the laws of physics.

Quote:
Originally Posted by dr_daw View Post
As for sample rate, I'd love some more insight on that one. Nyquist relates to the fact that the signal must be sampled a min of twice the highest frequency. That is only if you want to accurately reproduce the signal with MIN aliasing. There are benefits to oversampling. As we all know there are harmonics that can get missed during that principle of only 2xfrequency.
There are benefits to oversampling but they are mainly cost or processing benefits. Pretty much all ADCs initially sample at a single rate of many mHz and then decimate down to the user selected sample rate, which is the cheapest and most logical method of avoiding the need for a steep anti-alias filter in the analogue domain. There are also potential processing benefits from oversampling for certain non-linear processes, compression for example. Lastly, it is much easier (read cheaper) to implement relatively artefact free anti-alias and anti-imaging filters with a large transition band. A 96kHz sample frequency theoretically provides this large transition band and that is why some ADCs and some plugins may sound better at the higher sample rates. But now we are talking about how individual programmers and manufacturers implement the filters in their equipment/plugins rather than an intrinsic issue with the sample rate. So for example, you're more likely to hear a difference between 44.1kHz and 96kHz sampling rates in a budget ADC.

Harmonics beyond the range of human hearing are a red herring. Even if you have mics, mic pre-amps, power amps and speakers which can record or reproduce ultra-sonic frequencies your ear cannot respond to them anyway. The old chestnut of ultrasonic harmonics affecting the frequencies you can hear is also untrue, intermodulation can only occur in the ear between frequencies (harmonics) to which the ear is sensitive.

You asked what format I use; I usually record at 24bit 48kHz due to delivery requirements, very rarely I might record 96Khz (or higher) files for sound design purposes if I'm intending to pitch shift down an octave or more.

Quote:
Originally Posted by StewartFang View Post
I am not disputing your math or any math for that matter. I'm telling you that when you record bed tracks 32 bit sessions sound different than 24 bit sessions which also sound different than 16 bit sessions.

Th reasons why are unimportant. They just do buddy.
Nice try, we're not talking about the bit depth of the sessions, we're talking about the bit depth of the recording file format, which you know perfectly well!

Quote:
Originally Posted by StewartFang View Post
Why the heck should the other poster know about the math behind digital audio.
You say you don't dispute Top Jimmy's maths which means you accept the 24bit and 32bit files are essentially identical but then state you hear a difference between these two identical files?! This is a great example of why anyone who works with digital audio needs a basic understanding of digital audio: To avoid falling into the same trap as you and believing they can hear a difference between two identical files.

Quote:
Originally Posted by StewartFang View Post
Wait a minute now...Greg are you really saying that sample rates all sound the same too???
That's exactly what I'm saying, there is no audible difference between sample rates! However, there maybe an audible difference between how different manufacturers implement various sample rates. See my response to dr. daw above.

Quote:
Originally Posted by StewartFang View Post
Great engineers know how to recognize something special and have the ability to capture it...
Great engineers also recognise that perceptions can easily be biased/fooled but know enough about how audio and hearing work to help avoid jumping to impossible or ridiculous conclusions!

G
Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
24 or 32 bit float Bender macOS 26 03-13-2018 06:41 AM
32 Bit Float boogiemotel Pro Tools 10 28 01-09-2013 10:22 AM
new recorders run 24 bit vs 32 bit float rockguitarist1255 Pro Tools 10 1 01-07-2012 10:06 PM
32 bit Float Logichead Pro Tools 10 0 10-22-2011 11:08 AM
Help! What is a 32-bit Float Mix ????? Rhapsody 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Mac) 6 02-27-2010 09:46 AM


All times are GMT -7. The time now is 05:15 AM.


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