The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
32 bit AD/DA converter , where and why used it?!
Old 11th June 2019
  #31
Gear Addict
 
toofanstudio's Avatar
 

When I send my audio with 786 khz and 32 bit , do you think “ CABLE “ can be handling truly ?!
How about analog cable ?!
Old 11th June 2019
  #32
Lives for gear
 
IanBSC's Avatar
Quote:
Originally Posted by toofanstudio View Post
When I send my audio with 786 khz and 32 bit , do you think “ CABLE “ can be handling truly ?!
How about analog cable ?!
USB, ethernet and I2S can handle 2 channel 32/786.
Old 13th June 2019
  #33
Gear Addict
 
toofanstudio's Avatar
 

Quote:
Originally Posted by IanBSC View Post
USB, ethernet and I2S can handle 2 channel 32/786.
But , at last , for going to every analog gear , need to use analoge cable .
Old 13th June 2019
  #34
Gear Nut
 

32 bit/64 bit is great for internal daw processing.

but for ad-da, i'm not sure there is analog gear out there that can even come close to producing that kind of dynamic range... yet.
Old 13th June 2019
  #35
Gear Addict
 
toofanstudio's Avatar
 

Quote:
Originally Posted by gearstudent View Post
32 bit/64 bit is great for internal daw processing.

but for ad-da, i'm not sure there is analog gear out there that can even come close to producing that kind of dynamic range... yet.
Exactly, I think like you.
So how about cable ?!
About true energy of cable ?!
What is cable power for handling 32 bit or 192 khz or 180db dynamic range ?!?!
Just analog cable .
Because I know digital cable can do it .
Old 14th June 2019
  #36
Gear Guru
 
Muser's Avatar
looks like a smart way to get right to 32bit float. leaning on the preamps while improving their performance and features.

Old 4 weeks ago
  #37
Quote:
Originally Posted by Muser View Post
I noticed recently that I can set my macs audio setup for its own IO to 32bit float.
apparently it's hard to find out what the onboard converter specs are.

the 32bit float selection may have come in on High Sierra and if I keep it selected,
I can set the DAW to still record at 16bit for example, and it records a 16bit file.
but it doesn't force the Audio IO to change from 32bit float as it does so.
You’re confusing capture bit depth (fixed) with processing bit depth (floating point). Most daws now process at 64bit float, but capture is still 24butfuxed.


Quote:
Originally Posted by u87allen View Post
Actually, there’s no analog front end on the planet that has a low enough noise floor to take full advantage of all 32 bits. 32 bits works out to something like 194dB of dynamic range. Best I’ve heard of is in the 130dB range, which is like the 22 bit level.
This. Not worth recording at 32bit til we have a true 24b converter!
Old 4 weeks ago
  #38
Gear Guru
 
Muser's Avatar
Quote:
Originally Posted by psycho_monkey View Post
You’re confusing capture bit depth (fixed) with processing bit depth (floating point). Most daws now process at 64bit float, but capture is still 24butfuxed.
the version of Reaper is 64bit processing. so the case I’m explaining is not using any soundcard, just the Macs own IO. the standard IO in Audio Midi setup. it seems that in this case, the IO can be set to 32bit Float, Reaper can be set to access the Macs own on board IO and Reaper can be set to record at 16bit, while using the Macs IO, which remains set at 32bit Float. I got someone to also test this setting, but because they had Sierra, 32bit Float was not appearing as an option in Audio Midi setup for them. High Sierra seems to have included a lot more options in Audio Midi setup. I’d assume this is partly because High Sierra is a 64bit based OS.
Old 4 weeks ago
  #39
Quote:
Originally Posted by Muser View Post
the version of Reaper is 64bit processing. so the case I’m explaining is not using any soundcard, just the Macs own IO. the standard IO in Audio Midi setup. it seems that in this case, the IO can be set to 32bit Float, Reaper can be set to access the Macs own on board IO and Reaper can be set to record at 16bit, while using the Macs IO, which remains set at 32bit Float. I got someone to also test this setting, but because they had Sierra, 32bit Float was not appearing as an option. High Sierra seems to have included a lot more options in Audio Midi setup. I’d assume this is partly because High Sierra is a 64bit based OS.
64bit OS is to do with how the memory is addressed, and how much can be used by any one app. Not anything to do with audio bit depth.

"Setting the Mac IO to 32bit" doesn't make a lot of sense. The converter has to be fixed bit depth (since it can't move the decimal point for "live audio" if you see what I mean).

I don't really know what's happening if Reaper recording at 16b is capturing audio that's designated 32bit float coming from a fixed 24bit converter, but I suspect it's simply a 24bit stream being converted to 16b on they way in!
Old 4 weeks ago
  #40
Gear Guru
 
Muser's Avatar
Quote:
Originally Posted by psycho_monkey View Post
64bit OS is to do with how the memory is addressed, and how much can be used by any one app. Not anything to do with audio bit depth.

"Setting the Mac IO to 32bit" doesn't make a lot of sense. The converter has to be fixed bit depth (since it can't move the decimal point for "live audio" if you see what I mean).

I don't really know what's happening if Reaper recording at 16b is capturing audio that's designated 32bit float coming from a fixed 24bit converter, but I suspect it's simply a 24bit stream being converted to 16b on they way in!
right. I don't know what happening either. I'd expect it's a 24bit converter as well.
but I couldn't find anyone online who knew anything about the converters.
usually I notice that the Audio Midi setup changes if an applications preferences makes it change.
I noticed it when I started using Dante Via. I'm just really reporting a recent observation.
the inner workings of Macs tend to be obscure as apple never release schematics.
Old 4 weeks ago
  #41
Quote:
Originally Posted by Muser View Post
right. I don't know what happening either. I'd expect it's a 24bit converter as well.
but I couldn't find anyone online who knew anything about the converters.
usually I notice that the Audio Midi setup changes if an applications preferences makes it change.
I noticed it when I started using Dante Via. I'm just really reporting a recent observation.
the inner workings of Macs tend to be obscure as apple never release schematics.
It is, but of course it's a fairly lo-fi 24bit!
Old 4 weeks ago
  #42
Gear Addict
 
toofanstudio's Avatar
 

Quote:
Originally Posted by Muser View Post
right. I don't know what happening either. I'd expect it's a 24bit converter as well.
but I couldn't find anyone online who knew anything about the converters.
usually I notice that the Audio Midi setup changes if an applications preferences makes it change.
I noticed it when I started using Dante Via. I'm just really reporting a recent observation.
the inner workings of Macs tend to be obscure as apple never release schematics.
Analoge cable XLR - TRS and ... ‘s power is 20 bit.
At first , we should be change all of my cable to very high resolution cable like gold cable .
Old 4 weeks ago
  #43
Lives for gear
 
Brian Campbell's Avatar
 

Here on planet Earth the best you're going to get is 21 bits on a good day, the rest is just (thermal) noise. So what's the point?
Old 4 weeks ago
  #44
Lives for gear
 

So everyone......the new steinberg axr4 is in everyone's hands for either mac or (now) windows....is 32 bit integer.....and....well explain again what 32 bit integer means performance-wise on that specific interface.
Old 4 weeks ago
  #45
Lives for gear
 

Quote:
Originally Posted by thenoodle View Post
So everyone......the new steinberg axr4 is in everyone's hands for either mac or (now) windows....is 32 bit integer.....and....well explain again what 32 bit integer means performance-wise on that specific interface.
Yes. They are using the AKM AK5397EQ chip for the ADC.

Which also has these tech specs (in isolation):
THD+N: -108dB
DR, S/N: 127dB

And a dynamic range of only 119dB as used in the AXR4.

So does it even really matter that it is 32 bit int?
Old 4 weeks ago
  #46
Gear Guru
 
Muser's Avatar
Quote:
Originally Posted by psycho_monkey View Post
It is, but of course it's a fairly lo-fi 24bit!
I'd have some doubts about claims of it being the highest of fidelity. lets put it that way. but the IO is often a combo Analog/Toslink combo jack. so other Toslink converters can also be leveraged. but it is curious that I can set the basic IO to 32bit float on High Sierra for both in and out. I'd assume that if I connected a 16bit or a 24bit Toslink external converter, the core Audio would revert to the bit depth and sample rates optionally available for those converters. but I've also noticed that behavior when launching applications which are set to differing settings. in that I would find the settings in core Audio had changed.

as an example, when running Dante Via, Via sets up its own core Audio assignment which are always set to work at 48KHz and 32bit Float, when installed. It is independent of the normal core Audio IO. but the normal core Audio settings can be set at different rates to Via. even though the Via core Audio assignments can only work at 48KHz 32bit float and even though Via can route various applications own IO and various IO hardware, all at the same time. having reviewed descriptions from audinate responding to questions, audinate state that because of the way the technology works, they state that no sample rate conversion takes place. they didn't explain how that's the case however. I assume that would have required some in depth technical explanations.
Old 4 weeks ago
  #47
Gear Guru
 
Muser's Avatar
Quote:
Originally Posted by toofanstudio View Post
Analoge cable XLR - TRS and ... ‘s power is 20 bit.
At first , we should be change all of my cable to very high resolution cable like gold cable .
I’d assume the connector quality could be more important than the cable. because that’s where the most likely impedance barriers are going to occur. a well specified well constructed cable is not a ridiculous decision. though of course you can end up paying some ridiculous prices. also there are 75ohm and 110ohm analog cables for Digital. I think I remember Bob arguing they can sometimes be a good choice for analog Audio. I don’t personally have experience of anything about that though.
Old 4 weeks ago
  #48
Lives for gear
 

Quote:
Originally Posted by thenoodle View Post
So everyone......the new steinberg axr4 is in everyone's hands for either mac or (now) windows....is 32 bit integer.....and....well explain again what 32 bit integer means performance-wise on that specific interface.
That really doesn't explain much, nor it's any proof of justification for necessity of such step.

It's the similar thing like what happened at HiFi market years ago..
Converter chip vendors made newer lineups, which introduced direct support for I/O with higher bitdepths over I2S.
I was somewhat hand in hand with increased internal numeric precision used for decimation and interpolation filters in those chips.
I2S bus, used for outside communication with some other interfacing chips, MCUs or FPGA there, supports essentially arbitrary word lengths.

To me the main idea there wasn't to enable transmission for encoded noise lying 150dB under zero, but to streamline communication with some other chips doing DSP operations outside of DAC chips.. like when designer decide to implement own upsampling, decimation or say SRC elsewhere. So it can output samples with 32 bit precision and transfer those directly to or from converter chip without need for intermediate rounding or dithering.
Of course if you skip standard digital interfaces like AES, SPDIF, MADI.. and use "computer" external buses like USB, PCI, TB or have some network streaming, then you can rather easily pass those 32 bit samples directly to or from audio drivers and software.
Similarly like with uber high sample rates, where mentioned standard digital interfaces were ceiling of what's practically possible.

Of course, regardless of any end-to-end benefits, it's primarily great marketing item.. very simple, needs almost no explanation for majority of people.. the more, the better.
That spans from chip/IC vendors through audio gear makers, to software vendors. There is always simple bottom line.. you need to upgrade, because number gets higher and your current thingie is a weak link, because it doesn't support that.
I'm not angry about that, not at all, that's just how it works (mostly).. they all need to sell, in those markets isn't really anything like solved case/problem. Lot of times one company invent (or absurdly inflate) a problem, sell the solution.. and other competitors will follow that (to stay relevant). Other aspect there might be simple technical differentiation for flagship products.
I'd guess, that race never stops.. in few years from now, you certainly find bunch of people, who will be discussing products with megahertz PCM rates, running 64bit FP end to end and it's impact to Kind of Blue record, which would be sounding like Miles is standing right in front of them.

However lot of times it can happen, there can be also indirect technical benefits, because even when you don't use such new advertised rate or feature, new generation of chips might be better in other relevant aspects, surrounding analog circuitry might be further improved and tweaked to better accommodate "hyper" rates and depths.. etc. But of course that doesn't apply generally..

Sometimes people around me gets kinda confused by that and are for example surprised, that decades old well designed gear, which doesn't support those modes can clearly wipe the floor with some lower grade buzzword compliant converter/interface. Because everything has to be well balanced.. new IC doesn't save you from worse design of other parts. It's akin to putting multi megapixel tiny chip behind plastic lens in mobile phone camera.

As I've touched, you can see that most clearly at HiFi market, where is much larger portion of people following all trendy things.. but of course that applies also to semi-pro and pro studio market. Lately quite a few companies, who were rooted in pro audio further widen its reach towards HiFI.. (Mytek, Prism.. etc.).

So with all of that I'm not really surprised, that Steinberg made 384k/32bit audio interface. Even just for possible buzz it could create (Finally the first one in this segment.. WHOA!) among people, it's likely worth of that.
Chips were already here for long time (and other pro vendors also can use or already use, but didn't enabled those modes), bus to computer isn't bottleneck, they have own DAW.. hell, why not

Michal
Old 4 weeks ago
  #49
Gear Addict
 
toofanstudio's Avatar
 

Quote:
Originally Posted by Muser View Post
I’d assume the connector quality could be more important than the cable. because that’s where the most likely impedance barriers are going to occur. a well specified well constructed cable is not a ridiculous decision. though of course you can end up paying some ridiculous prices. also there are 75ohm and 110ohm analog cables for Digital. I think I remember Bob arguing they can sometimes be a good choice for analog Audio. I don’t personally have experience of anything about that though.
Cable and connector both of them are important .
I haven’t any good sources for impedance of cable , that which number good to which part of sound . If you know please send some sources.!
Old 4 weeks ago
  #50
Lives for gear
 

Quote:
Originally Posted by msmucr View Post
Quote:
Originally Posted by thenoodle View Post
So everyone......the new steinberg axr4 is in everyone's hands for either mac or (now) windows....is 32 bit integer.....and....well explain again what 32 bit integer means performance-wise on that specific interface.
I'm not really surprised, that Steinberg made 384k/32bit audio interface. Even just for possible buzz it could create (Finally the first one in this segment.. WHOA!) among people, it's likely worth of that.
Chips were already here for long time (and other pro vendors also can use or already use, but didn't enabled those modes), bus to computer isn't bottleneck, they have own DAW.. hell, why not

Michal
So.....it's bs marketing? The akm is 32bit integer, therefore Steinberg is stating that the interface is 32-bit integer but....there is absolutely zero going on in the interface 32-bit-integer-wise that benefits a human on planet earth who may buy because of that listed spec?

As this thread specifically pertains to incoming interfaces with the spec (despite some people getting confused)...I'd just like to be sure some real-life benefit hasn't been overlooked here.

I specifically plan to point Steinberg to this if no valid benefit comes up.
Old 4 weeks ago
  #51
Lives for gear
 

It is marketing.
They aren't even up to 24 bit integer dynamic range which is 144dB.
Far, far from it.
Old 4 weeks ago
  #52
Gear Guru
 
Muser's Avatar
Quote:
Originally Posted by toofanstudio View Post
Cable and connector both of them are important .
I haven’t any good sources for impedance of cable , that which number good to which part of sound . If you know please send some sources.!
changes in medium cause changes in impedance. so if you have connectors with poor tolerance you could get variables in contact and if that includes air gaps you have two different mediums. air and metal. those two mediums present different impedances to the electrical signal. so a good connector have good tolerances so that the connections are, well made.

gold is good in that it is soft, so it can tend to grip well and it also tends to be less resistant to tarnishing. switchcraft are good connectors for in situ connections and Neutric are good for cables which are connected and disconnected a lot in intensive applications. because they have good strain relief. van damme and Hosa are perfectly fine cables and don't tend to demand premium prices.
Old 4 weeks ago
  #53
Lives for gear
 

Quote:
Originally Posted by thenoodle View Post
So.....it's bs marketing? The akm is 32bit integer, therefore Steinberg is stating that the interface is 32-bit integer but....there is absolutely zero going on in the interface 32-bit-integer-wise that benefits a human on planet earth who may buy because of that listed spec?

As this thread specifically pertains to incoming interfaces with the spec (despite some people getting confused)...I'd just like to be sure some real-life benefit hasn't been overlooked here.

I specifically plan to point Steinberg to this if no valid benefit comes up.
It's definitely usable for marketing, as I've expanded before. Also it's not much effort nowadays to simply enable high rates and longer wordlengths from chips and open data paths to driver, especially if one vendor can control everything from hardware up to DAW.
With regards to real world benefits, I'm personally not convinced and don't think, it's needed for capture and playback of audio. Physical limits of dynamic range for analog electronics (primarily due to Johnson noise), not to mention recording or listening environments changed a bit.
So when all other aspects will be equal (eg. you'll be using the same analog source with the same interface) and you'd do two captures in 32 and 24 bits to your DAW (if it supports that), then I'd bet, noisefloor won't change, simply because bitdepth isn't any bottleneck there.
Similarly for straight playback of 32 or 24 bit PCM audio, real-world boundaries are elsewhere.

Precision for processing in chip itself are bit different thing.. If you for example imagine, what happens in modern DA converter chip, then there is reconstruction filter before oversampling, conversion to multi-bit SDM signal (that's bit different from 1-bit DSD) using modulators and possibly also digital level control. For all of that, increased internal precision and longer accumulators makes sense, when you aim to minimize errors for calculations there.

As I've touched before, if you decide to do some functionality outside of the converter chip, then you can use high rates and theoretically also benefit something with higher bitdepths.
For example ESS DAC chips can be fed with 32bit PCM audio up to 1.5 MHz, AKM up to 768 kHz or high rate DSD.
At HiFi market you can find, some players/oversamplers, which can fed downstream DAC with very high rates and larger bitdepths. Essentially bypassing internal DAC oversampling or moving it out of picture (because it's then working at very high sample rates far from audioband), or convert PCM to DSD for playback (both ends at multi-bit SDM).. etc.
So user can then chose various different oversampling algorithms according his taste and for instance let it eat whole i7 CPU just with that in real-time
Again, I'm not convinced, it's really worth of that.. but you know, for apparently lot of people out there, more intricate, expensive and convoluted setup, more magic.. so hassle or inconvenience can the least problematic thing. And say 32 bit and 768k via USB can be really good argument for them to upgrade, because it opens another way for tweaking .

But all of that rarely applies to pro audio.
If there is some offloading of oversampling, decimation and related filtering from the chip or ASRC, it's typically processed "under the hood" at hardware FPGA in converter/interface (like Benchmark, Cranesong.. etc.) and not at some additional software running at DAW.
So it's transparent for user, and from his point, hardware is working with "normal" rates and bitdepths..

Only thing, where I was intrigued about high rates for my personal use, is tape restoration, if some company would make affordable (or rentable) Plangent like processing for wow and flutter removal, then converter running at 384k (say like RME ADI-2 Pro or that Steinberg) with good parameters could reach to ultrasonic range and capture bias tones, which are used basically as FM carrier for the restoration process.. but that's of course super niche and doesn't have any relation to 32 bit PCM I/O.

So I'm curious, what Steinberg possibly tell you, if you ask them..
I'd guess, it would be something like, it's for more precise sound and you can always try, whether it would be working for your material

Michal
Old 3 weeks ago
  #54
Gear Head
 
Bodhifile's Avatar
It seems to me that a relevant benefit of 32-bit converters is to avoid multiple dithering instances during mixing/mastering. You can avoid dithering on every bounce or outboard AD/DA loop within a project. Most DAW's have the option to disable automatic internal dithering. I think I read somewhere that 64-bit to 32-bit truncation distortion is almost irrelevant (though there is also airwindows ditherfloat if inclined). If staying 32-bit post recording- A or D, only one dithering instance is needed (recommended) when converting final mixdown to 24 or 16-bit.
Old 3 weeks ago
  #55
Lives for gear
 

Quote:
Originally Posted by Bodhifile View Post
It seems to me that a relevant benefit of 32-bit converters is to avoid multiple dithering instances during mixing/mastering. You can avoid dithering on every bounce or outboard AD/DA loop within a project. Most DAW's have the option to disable automatic internal dithering. I think I read somewhere that 64-bit to 32-bit truncation distortion is almost irrelevant (though there is also airwindows ditherfloat if inclined). If staying 32-bit post recording- A or D, only one dithering instance is needed (recommended) when converting final mixdown to 24 or 16-bit.
You are confusing two type of numbers.

This is a 32 bit integer format for the AD/DA

The engine of DAWs and such are 64 bit float. The files generated are 32 bit float from DAWs (like Pro Tools or Cubase, etc.) Not 32 bit integer. Sending out from a DAW 32 bit float to a 32 bit integer converter will still need dithering (and type conversion)

The dithering question is still in play (though pointless) and you aren't avoiding it with 32 bit integer converters (like the AKM chip Steinberg is using.)
Old 3 weeks ago
  #56
Gear Head
 
Bodhifile's Avatar
Quote:
Originally Posted by pentagon View Post
You are confusing two type of numbers.

This is a 32 bit integer format for the AD/DA

The engine of DAWs and such are 64 bit float. The files generated are 32 bit float from DAWs (like Pro Tools or Cubase, etc.) Not 32 bit integer. Sending out from a DAW 32 bit float to a 32 bit integer converter will still need dithering (and type conversion)

The dithering question is still in play (though pointless) and you aren't avoiding it with 32 bit integer converters (like the AKM chip Steinberg is using.)
Thanks for the response. I am aware that integer and floating point are different, not too up on the differences though. So let me ask you this: pointless in your opinion to avoid dithering when recording a stem at 32-bit floating point through a 24-bit AD converter? Thanks
Old 3 weeks ago
  #57
Lives for gear
 

Yes - that would be pointless but never happens.

You are never dithering when recording an input from a 24 bit integer AD coming in to a 32 bit float file. All 24 of those bits can be fully represented in 32 float. The point of dithering is when there is some form of truncation; that is not happening in this scenario.

If the discussion is about going out from 32 bit float file to a 24 bit DA, that would be something to discuss. That is where the discussion is happening (ad nauseam) since there is a form of truncation. Here I think it is pointless as it’s been shown the noise created is way below human hearing (and sound reproduction systems.) it becomes an educational exercise. The airwindows float is about/for plugin makers who go between DAW engine and file where there can be some influence in cumulative error.
Old 3 weeks ago
  #58
Gear Head
 
Bodhifile's Avatar
yeah, I meant an AD/DA loopback stem scenario: 32/64 bit float mix> 24-bit DA>24-bit AD> 32-bit float rec. Staying 32-bit float post initial recording at 24-bit sounds better to me, mainly for bouncing stems inside the project etc to avoid excessive dithering or excessive truncation (if doing a lot). Haven't come to a definitive conclusion as to whether recording stems or mixdowns OTB with or without dither/24-bit or 32-bit is "best" though. I'm sure most would say it's unnecessary to record back at 32-bit.
Old 3 weeks ago
  #59
Lives for gear
 

“Excessive” dithering (non-noise shaped) is a myth. There’s a very good dithering discussion in the Mastering forum if want to see why (also about float dithering/truncation).
It gets technical if that’s not your thing.

I’ll leave it at that since this is getting way off topic. You can go snoop in that thread if it interests you.
Old 3 weeks ago
  #60
Lives for gear
We just love those numbers.
Even if they say absolutely nothing.
https://en.m.wikipedia.org/wiki/Johnson–Nyquist_noise
Post Reply

Welcome to the Gearslutz 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…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump