The No.1 Website for Pro Audio
32 Bit A/D D/A Converter Recorder
Old 17th February 2012
  #1
Here for the gear
 
🎧 10 years
32 Bit A/D D/A Converter Recorder

Is there such a thing that can record (via studio box, portable recorder, sound card) 32bit sound files? Im thinking 32 bit 192khz or even 32 bit 384khz. I've seen some software programs offer 32 bit files as an option for editing, recording, etc. So how are people recording these 32 bit files?

My only option now is to use a Kord MR-1000, but I was hoping for an official unit that supports the spec. Basically something that supports DXD or better.

I've looked for sound cards and they all seem to stop at 24bit/192khz. Same with studio boxes. Any ideas?
Old 17th February 2012
  #2
pmx
Gear Addict
 
pmx's Avatar
 
🎧 10 years
32 is 24 bits information with 8 bits floating. try reading up on bit depth, that should explain a lot of things. 32 bit converters do not exist.
Old 17th February 2012
  #3
Lives for gear
 
1 Review written
🎧 15 years
Outside the realm of marketing snakeoil the short answer is no.

A longer answer involves pointing to reality of what's involved in translating one form of energy, pressure waves for example, into another, electro-magnetic, etc.

At this time, by necessity, digitizing pressure waves is an integer process. Current DACs currently are hard pressed to utilize theoretic 20 bit limits. 24 bit is used for software/firmware coding efficiency not because anyone can distinguish, via ear, between 20 & 24 bit depth. There are thermodynamic reasons why pressing the limits of 24 bit ADC is impractical. Software applications use 32 bit floating point math (or floating point math in general) for greater precision, reducing 'rounding' errors (which accumulate) that are an inevitable result of integer math, in calculations.
Old 17th February 2012
  #4
Lives for gear
 
lpkyer's Avatar
 
🎧 15 years
Black Lion Audio is coming up with a 32 bits fixed / 384khz recorder pretty soon.
Old 17th February 2012 | Show parent
  #5
Lives for gear
 
5 Reviews written
🎧 15 years
Quote:
Originally Posted by lpkyer ➡️
Black Lion Audio is coming up with a 32 bits fixed / 384khz recorder pretty soon.
are you serious?
Old 17th February 2012 | Show parent
  #6
Lives for gear
 
5 Reviews written
🎧 15 years
Quote:
Originally Posted by pmx ➡️
32 bit converters do not exist.
there are several
Old 17th February 2012
  #7
70% Coffee, 30% Beer
 
Doc Mixwell's Avatar
 
1 Review written
🎧 15 years
hey braxus,

are you close mic'n an orchestra of jackhammers?

In a bat cave?

What is the use of all that dynamic range and frequency content for your application?

If I might ask you the question,
Old 17th February 2012
  #8
Lives for gear
 
Joe Porto's Avatar
 
🎧 15 years
There is a 28bit converter, that can record anything from the level of a dynamic mic up to above line level.

DSD and related formats (Korg MR1000) are 1 bit. Totally different process....

24 bit converters are already limited by the dynamic range of their analog circuitry, so I'm not sure what advantage a higher bit rate will give you for recording purposes.
Old 17th February 2012 | Show parent
  #9
Lives for gear
 
lpkyer's Avatar
 
🎧 15 years
Quote:
Originally Posted by Joe Porto ➡️
24 bit converters are already limited by the dynamic range of their analog circuitry, so I'm not sure what advantage a higher bit rate will give you for recording purposes.
More precision in the quantification process.


Quote:
Originally Posted by robertshaw
are you serious?
yes ! check them out on facebook.
Old 18th February 2012 | Show parent
  #10
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by lpkyer ➡️
More precision in the quantification process.
How so? Each bit still represents 6dB of audio. Adding more bits won't change this. How does it make it more precise? The addition of another 8 bits for 24bit a/d conversion simply means you capture 48dB more signal... but in the case of digital converters... it will just be johnson nyquist noise.
Old 18th February 2012
  #11
Lives for gear
 
Plush's Avatar
 
5 Reviews written
🎧 15 years
Put down your pipes---

Please read the Bible:

John Watkinson: The Art of Digital Audio
(Focal Press)

Then you will be able to more confidently discuss the issue.

The inherent noise level of analog electronics blinds off before reaching even the 24 bit level.
Old 18th February 2012
  #12
Lives for gear
 
Demonslave's Avatar
 
🎧 10 years
Quote:
Originally Posted by Roc Mixwell ➡️
hey braxus,

are you close mic'n an orchestra of jackhammers?

In a bat cave?

What is the use of all that dynamic range and frequency content for your application?

If I might ask you the question,
Hahaha
Old 18th February 2012 | Show parent
  #13
Lives for gear
 
lpkyer's Avatar
 
🎧 15 years
Quote:
Originally Posted by BasketCase Audio ➡️
How so? Each bit still represents 6dB of audio. Adding more bits won't change this. How does it make it more precise? The addition of another 8 bits for 24bit a/d conversion simply means you capture 48dB more signal... but in the case of digital converters... it will just be johnson nyquist noise.
No more oversampling for lower bits !
Old 18th February 2012 | Show parent
  #14
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by lpkyer ➡️
No more oversampling for lower bits !

Lower bits? You mean the last 24dB that are well out of the range of human hearing and submerged within the inherent noise of the analogue circuitry that comes before and after the conversion itself?
Old 18th February 2012
  #15
Gear Maniac
 
avebr's Avatar
 
🎧 10 years
Stagetecs' TrueMatch converters (direct mic input AD) claimed 28bit resolution. it is at least a 10 year old technology.

Current XMIC+ Microphone-input Boards claims 32bit, but it "could have been achieved using only 27 bits. One advantage of Stage Tec’s strategy of using the 32-bit conversion method on the XMIC+ board is that the intrinsic noise acts as a »natural« dither, which moves the quantisation noise far below the noise level of the connected source". It's rated Dynamic range: 158 dB(A) (typ.) @ 24 dBu

Such higher resolution bits can only exists inside their proprietary connections/machines (Nexus system).

If it gets outputed to the external world thru a common digital connection (like aes,spdif,adat,madi,...), it will be 'reduced' (truncated I believe) to 24bit.

Those are the highest bit pcm converters I've known. But only on their proprietary world.


all the best,
ave.
Old 18th February 2012 | Show parent
  #16
Gear Head
 
🎧 10 years
Quote:
Originally Posted by BasketCase Audio ➡️
How so? Each bit still represents 6dB of audio. Adding more bits won't change this. How does it make it more precise? The addition of another 8 bits for 24bit a/d conversion simply means you capture 48dB more signal... but in the case of digital converters... it will just be johnson nyquist noise.
I'm no expert, but since dB is relative, you ARE getting more precise. Recording at 24 bits vs 16 bits doesn't mean your recording will be any louder. Instead, it has more bits of precision at the same loudness.

To address the 1st post, the Axcel2 resynthesizer can sample at 384KHz (link below). All this high sample rates seem ridiculous and wasteful since higher sample rates mean higher frequencies, and of course humans can't hear those extra 1,2 and 3 octaves these higher rates give you.

Acxel-en_hw
Old 18th February 2012 | Show parent
  #17
Lives for gear
 
Joe Porto's Avatar
 
🎧 15 years
Quote:
Originally Posted by Elhardt ➡️
I'm no expert, but since dB is relative, you ARE getting more precise. Recording at 24 bits vs 16 bits doesn't mean your recording will be any louder. Instead, it has more bits of precision at the same loudness.
The "precision" is not distributed throughout the dynamic range. It's added to the bottom. Once you go below the analog noise floor, any additional "precision" is being masked by analog hash.
Old 19th February 2012
  #18
restpause
Guest
Perhaps the extra dynamic range *IS* added at the top of the amplitude scale? This could imply such converters are not for typical consumer/prosumer/pro audio use, but for high (perhaps SPL) industrial uses instead. Rock Mixwell said it best ^^^^.
Old 19th February 2012 | Show parent
  #19
Gear Head
 
🎧 10 years
Quote:
Originally Posted by Joe Porto ➡️
The "precision" is not distributed throughout the dynamic range. It's added to the bottom. Once you go below the analog noise floor, any additional "precision" is being masked by analog hash.
Well, yes, the more bits of precession, the more they're added to the bottom as far as dB goes as you say. If visually viewing the waveform they're distributed evenly throughout. I guess the point I was trying to make was that 24 bits doesn't get you louder sound, it gets you sound with more precession even if a couple of bits are below the noise floor, though that doesn't apply to digitally generated waveforms.
Old 19th February 2012 | Show parent
  #20
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by Elhardt ➡️
Well, yes, the more bits of precession, the more they're added to the bottom as far as dB goes as you say. If visually viewing the waveform they're distributed evenly throughout. I guess the point I was trying to make was that 24 bits doesn't get you louder sound, it gets you sound with more precession even if a couple of bits are below the noise floor, though that doesn't apply to digitally generated waveforms.

There isn't anything more precise about 24bit. The dB scale in digital audio is absolute. 1 bit equals 6dB for both 16 and 24bit. You can record a signal at -50dB FS in your 24 bit system and effectively output the same data as though you were recording at -2dB FS within a 16bit system. The only difference is the leading 8 bits beyond the signalling bit for the two's compliment system are padded out with 0's. Both would have comparable noise and quantisation errors.

"Louder" on the other hand, is relative. Both 16 and 24bit systems are inversely scaled, starting at 0dB FS at the top and going down. If you consider the difference between the degree of soft and loud that each respective system is capable of, 24bit certainly is "louder".

Consider it like computer screens. Higher resolution doesn't mean more precision. A computer with a 2400x1200 display has a higher resolution than one with a 1200x600 display... but if they are both displaying 100 dots per inch, neither is "more precise" than the other. One just has an image that is twice the area. The same applies for digital audio. 24bit is higher resolution, but it isn't more precise.
Old 19th February 2012 | Show parent
  #21
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by BasketCase Audio ➡️
There isn't anything more precise about 24bit. The dB scale in digital audio is absolute. 1 bit equals 6dB for both 16 and 24bit. You can record a signal at -50dB FS in your 24 bit system and effectively output the same data as though you were recording at -2dB FS within a 16bit system. The only difference is the leading 8 bits beyond the signalling bit for the two's compliment system are padded out with 0's. Both would have comparable noise and quantisation errors.

"Louder" on the other hand, is relative. Both 16 and 24bit systems are inversely scaled, starting at 0dB FS at the top and going down. If you consider the difference between the degree of soft and loud that each respective system is capable of, 24bit certainly is "louder".

Consider it like computer screens. Higher resolution doesn't mean more precision. A computer with a 2400x1200 display has a higher resolution than one with a 1200x600 display... but if they are both displaying 100 dots per inch, neither is "more precise" than the other. One just has an image that is twice the area. The same applies for digital audio. 24bit is higher resolution, but it isn't more precise.
Not so.

Your first example is true in theoretical, mathematical world. In real life it is not. Reason: the "top bits" in 16 bit and 24 bit system are "clean", the system noise is at the least significant bits. In a good 16 bit system 15 largest bits are more or less perfect, but the least significant bit is more or less random in nature. In the very best 24 bit real life systems 21 largest bits are "perfect", 3 least significant bits are random, as there is enough system noise at those minuscule voltage levels (2^8 smaller than what 16 bit system sees) and approach the thermal noise of the system. So, if you take your -50 dBFS 24 bit signal and -2dBFS 16 bit signal, only the top 13 bits will match, below that it is noise, in 24 bit signal first.

Second paragraph: 24 bit system is not louder when using conventional DACs, which have a fixed output level for 0 dBFS. It can go softer, though, without noise. You can play it louder with the same noise floor, but only by adding analog gain post DAC, which is another matter all together. You can play anything loud or soft by turning the big volume knob in the amplifier.

Third paragraph: Comparing audio to photographs/displays does not work, because pictures can be enlarged, audio can not. 24 bit signal is more precise, as it can describe smaller variations in the signal, theoretically 1/256 as small as 16 bit system, 1/32 in real life (16 versus 21 real bits).
Old 19th February 2012 | Show parent
  #22
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by Petrus ➡️
Not so.

Your first example is true in theoretical, mathematical world. In real life it is not. Reason: the "top bits" in 16 bit and 24 bit system are "clean", the system noise is at the least significant bits. In a good 16 bit system 15 largest bits are more or less perfect, but the least significant bit is more or less random in nature. In the very best 24 bit real life systems 21 largest bits are "perfect", 3 least significant bits are random, as there is enough system noise at those minuscule voltage levels (2^8 smaller than what 16 bit system sees) and approach the thermal noise of the system. So, if you take your -50 dBFS 24 bit signal and -2dBFS 16 bit signal, only the top 13 bits will match, below that it is noise, in 24 bit signal first.

Second paragraph: 24 bit system is not louder when using conventional DACs, which have a fixed output level for 0 dBFS. It can go softer, though, without noise. You can play it louder with the same noise floor, but only by adding analog gain post DAC, which is another matter all together. You can play anything loud or soft by turning the big volume knob in the amplifier.

Third paragraph: Comparing audio to photographs/displays does not work, because pictures can be enlarged, audio can not. 24 bit signal is more precise, as it can describe smaller variations in the signal, theoretically 1/256 as small as 16 bit system, 1/32 in real life (16 versus 21 real bits).

Argh, yes. You are right. Its been a while since I studied this stuff. Truncation happens from the least significant bit first, so indeed. For some reason I had it around the other way...

My apologies.
Old 19th February 2012 | Show parent
  #23
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by avebr ➡️
Stagetecs' TrueMatch converters (direct mic input AD) claimed 28bit resolution. it is at least a 10 year old technology.
Yes but they use four standard 24bit AD's in parallel with different analog input gain and smart DSP post AD's to put the pieces together.

I see no benefit to that process compared to a well trimmed standard solution (mic pre + AD with proper gain setting) in most situations.

The only real life benefit is that you don't have to care about setting your levels at the pre. Just plug your mic's and press rec.


/Peter
Old 19th February 2012
  #24
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by braxus ➡️
Is there such a thing that can record (via studio box, portable recorder, sound card) 32bit sound files? Im thinking 32 bit 192khz or even 32 bit 384khz. I've seen some software programs offer 32 bit files as an option for editing, recording, etc. So how are people recording these 32 bit files?

My only option now is to use a Kord MR-1000, but I was hoping for an official unit that supports the spec. Basically something that supports DXD or better.

I've looked for sound cards and they all seem to stop at 24bit/192khz. Same with studio boxes. Any ideas?
It's all about noise floor and bandwith. First you need to consider the release format for your work and then you need to consider what kind of set up this material will be played at. The drawback of small wordlengths/few bits in digital audio is that the noise floor is increased. With 24bit audio you will never hear the noise (if the converters used are top notch) unless you crank the volume way above normal settings.

For most listeners and material 16bit is fine. Ther are exceptions though and it's when you deal with dynamic material (not very common) and you have beefy amps with relatively high sensitivity speakers.

Capturing and playing back audio with 24bit sampling is more than enough but you still want to use higher rez while processing in your DAW. By increasing teh wordlength to 32-64bit or so you can do your EQ, mixing and stuff without introducing significant errors to the end result. You don't need to capture at 64bit in order to make good use of a 64bit audio engine in your DAW.


/Peter
Old 21st February 2012 | Show parent
  #25
Gear Head
 
🎧 10 years
Quote:
Originally Posted by BasketCase Audio ➡️
Consider it like computer screens. Higher resolution doesn't mean more precision. A computer with a 2400x1200 display has a higher resolution than one with a 1200x600 display... but if they are both displaying 100 dots per inch, neither is "more precise" than the other. One just has an image that is twice the area. The same applies for digital audio. 24bit is higher resolution, but it isn't more precise.
I think somebody else already cleared this up, but just to hammer it home: going to a higher resolution display but keeping the physical screen the same size does increase precision. You would have more pixels to represent a portion of the screen and so more details can be seen. As I said earlier, people aren't recording in 24 bits for louder sound.
Old 21st February 2012
  #26
Lives for gear
 
frans's Avatar
 
10 Reviews written
🎧 15 years
In case the engineering side of a "true" 32bit resolution recording would be solved the advantage would be in the details of the parts of the signal that are lower in volume. If you start with 32bit fullscale (=peaks) as your (pick a standard you are comfortable with, perhaps broadcast line levels) upper end you will push the shortcomings of PCM* audio coding into the noisefloor. Then if DAW companies would allow their software to actually have headroom instead of insisting 0dB to sit at the edge of clipping, digital would be more foolproof. We could have "0dB" at -18 (just as an example, okay?) and the upper rest would be headroom, just like in the old days. Then there still would be lots of resolution in the range below. Digital black would sit deep, deep on the bottom of the signal, way below the noisefloor, no more smear in silent signal tails.

If you instead count the upper 8 bits of 32 bits on top of the 24 bits we have now, you got it all wrong and there really would be no use for this, having it all upside down.

*PCM = pulse code modulation, read Dan Lavrys whitepaper if you already haven't
Lavry Engineering
Old 21st February 2012 | Show parent
  #27
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by frans ➡️
Then if DAW companies would allow their software to actually have headroom instead of insisting 0dB to sit at the edge of clipping, digital would be more foolproof. We could have "0dB" at -18 (just as an example, okay?) and the upper rest would be headroom, just like in the old days.
In a few days pop/rock producers and mastering engineers would start pushing peaks to +18 dBFS, and compress the minimum signal level to + 14 dBFS and we would be in square one...
Old 21st February 2012 | Show parent
  #28
Lives for gear
 
frans's Avatar
 
10 Reviews written
🎧 15 years
Quote:
Originally Posted by Petrus ➡️
In a few days pop/rock producers and mastering engineers would start pushing peaks to +18 dBFS, and compress the minimum signal level to + 14 dBFS and we would be in square one...
DAW companies could force a "turn it down, dumbass!" voiceover in the mix if levels exceed 0dB ...
Old 21st February 2012
  #29
Gear Nut
 
🎧 10 years
32 bits fixed / 384khz recorder ?

can someone explain why we need that ?
in time where all the music is cheap mp3
Old 21st February 2012 | Show parent
  #30
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by 24track ➡️
32 bits fixed / 384khz recorder ?

can someone explain why we need that ?
in time where all the music is cheap mp3
Do not worry, it is impossible to make anyway, unless the rail voltages in mics, preamps and converters are raised to over 10000 V. With that sort of power units it would hardly be portable, so it would certainly stay fixed!
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 60 views: 16059
Avatar for smokingrobot
smokingrobot 29th May 2012
replies: 8304 views: 772421
Avatar for Synthpark
Synthpark 1 hour ago
replies: 9 views: 469
Avatar for caronte23
caronte23 27th May 2022
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump