Jump to content
IGNORED

32 bit sound


Recommended Posts

Hi, and apologies if this has been discussed extensively before but I could not find the right answers.

 

Here's the issue: I know that industry and download sites push 24 bit digital audio and, since CES this year, perhaps also DSD, but what about 32 bit?

 

I thought about it during the summer holidays and now started digitalizing my vinyl in 32 bit float, using an RME Fireface UCX, connected via Firewire cable to a Macbook Pro, with Vinyl Studio as recording software.

 

For playback I use Audirvana, which allows 32bit playback, as - so it seems - does iTunes. As DAC am using a Hegel HD25: it contains 32 bit converters but, the company informs me, it actually cannot receive 32 bit files. So the DAC is limited to receiving 24 bits, which are then upsampled to 32 bit. But when the DAC is connected via USB and 'seen' by the Mac, AudioMidi Setup identifies it as 32 bit, not 24 bit. Can someone explain what is going on here? Does the Mac with iTunes and Audirvana really allow 32 bit playback, or is everything automatically downsampled to 24 bit before leaving the computer?

 

I need to add that, irrespective of what is going on technically, the improvements are evident. I have now done a few test recordings in both 24 and 32 bit float and can hear a clear improvement when playing back 32 bit songs, particularly with the harmonics of lower-level instruments and voices and in terms of image depth. 32 bit sound more three-dimensional and 'analogue-like' than 24 bit.

 

Thanks for any explanations that you might have.

Link to comment

Hi tresaino,

 

Hi, and apologies if this has been discussed extensively before but I could not find the right answers.

 

Here's the issue: I know that industry and download sites push 24 bit digital audio and, since CES this year, perhaps also DSD, but what about 32 bit?

 

I thought about it during the summer holidays and now started digitalizing my vinyl in 32 bit float, using an RME Fireface UCX, connected via Firewire cable to a Macbook Pro, with Vinyl Studio as recording software.

 

For playback I use Audirvana, which allows 32bit playback, as - so it seems - does iTunes. As DAC am using a Hegel HD25: it contains 32 bit converters but, the company informs me, it actually cannot receive 32 bit files. So the DAC is limited to receiving 24 bits, which are then upsampled to 32 bit. But when the DAC is connected via USB and 'seen' by the Mac, AudioMidi Setup identifies it as 32 bit, not 24 bit. Can someone explain what is going on here? Does the Mac with iTunes and Audirvana really allow 32 bit playback, or is everything automatically downsampled to 24 bit before leaving the computer?

 

I need to add that, irrespective of what is going on technically, the improvements are evident. I have now done a few test recordings in both 24 and 32 bit float and can hear a clear improvement when playing back 32 bit songs, particularly with the harmonics of lower-level instruments and voices and in terms of image depth. 32 bit sound more three-dimensional and 'analogue-like' than 24 bit.

 

Thanks for any explanations that you might have.

 

First and foremost, do *not* get fooled by the numbers game. Every week it seems a new DAC is presented to the market where the marketers have upped the numbers, there are claims for 32-bits, there are claims for 384k and 768k. In my experience, the reality is that there is not a lot of gear that can actually deliver 24-bits cleanly, all the way to the lowest order bits. A great deal of hardware spec'd for 192k performs *worse* at rates like this than it does at lower rates like 96k. (What does this say about claims for even higher numbers?)

 

Based on the audible results, the very significant increase in demands the 4x rates (like 192k) place on the designer for accurate clocking and for analog stage performance at wide bandwidth are apparently not so easy to meet, based on the audible results. Some will go so far as to publish "white papers" claiming rates higher than 2x are not a good idea. (Funny though, how those who create products fully capable of demonstrating the potential of 4x rates, don't make such claims.)

So, watch out for the "DAC of the week" syndrome.

 

Another point: 32-bit float *is* 24-bit audio. It is quite useful for the internal math utilized for processing a digital audio signal. Think of it as a 24-bit signal, temporarily allowed the to move around within the larger area encompassed by the 32-bit "package". This adds headroom for processing. But the audio is still 24-bits.

(This should not be confused however, with some of the marketing claims.)

The software I use for recording and mastering has data paths that are 80-bits wide (i.e., it performs its internal math to a precision of 80-bits) but the audio I'm listening to while it does this, is 24-bits.

 

Also, taking a 24-bit signal to 32-bits (or 32-bit float, which is not the same thing) involves padding the low order bits with zeros. This should not be confused with upsampling or resampling, which is changing the sample rate. Similarly, truncating the low order bits to take a 32-bit file to 24-bits is not downsampling (or resampling downward either).

Sampling rate defines how many times the analog signal is "looked at" per second, each "look" becoming a single digital "word" (also referred to as a "sample"). If we convert the sampling rate (up or down), the word length remains the same. (Converting sample rate and changing word length are two different processes.)

 

Word length (sometimes referred to as "bit depth") describes how many digital bits are used in each digital "word". So, 16-bit audio at 44.1 kHz sampling rate (the CD standard) will contain 44,100 "samples" (or "words") per second and each one is 16-bits long (e.g., 1011010111110001). If we padded the example in the previous sentence to 24-bits, it might look like this: 101101011111000100000000. Once we processed that 24-bit signal, the sample in the previous sentence might look like this: 101101011111000101011001.

 

Hope this helps.

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

Link to comment
Hi tresaino,

 

 

 

First and foremost, do *not* get fooled by the numbers game. Every week it seems a new DAC is presented to the market where the marketers have upped the numbers, there are claims for 32-bits, there are claims for 384k and 768k. In my experience, the reality is that there is not a lot of gear that can actually deliver 24-bits cleanly, all the way to the lowest order bits. A great deal of hardware spec'd for 192k performs *worse* at rates like this than it does at lower rates like 96k. (What does this say about claims for even higher numbers?)

 

Based on the audible results, the very significant increase in demands the 4x rates (like 192k) place on the designer for accurate clocking and for analog stage performance at wide bandwidth are apparently not so easy to meet, based on the audible results. Some will go so far as to publish "white papers" claiming rates higher than 2x are not a good idea. (Funny though, how those who create products fully capable of demonstrating the potential of 4x rates, don't make such claims.)

So, watch out for the "DAC of the week" syndrome.

 

Another point: 32-bit float *is* 24-bit audio. It is quite useful for the internal math utilized for processing a digital audio signal. Think of it as a 24-bit signal, temporarily allowed the to move around within the larger area encompassed by the 32-bit "package". This adds headroom for processing. But the audio is still 24-bits.

(This should not be confused however, with some of the marketing claims.)

The software I use for recording and mastering has data paths that are 80-bits wide (i.e., it performs its internal math to a precision of 80-bits) but the audio I'm listening to while it does this, is 24-bits.

 

Also, taking a 24-bit signal to 32-bits (or 32-bit float, which is not the same thing) involves padding the low order bits with zeros. This should not be confused with upsampling or resampling, which is changing the sample rate. Similarly, truncating the low order bits to take a 32-bit file to 24-bits is not downsampling (or resampling downward either).

Sampling rate defines how many times the analog signal is "looked at" per second, each "look" becoming a single digital "word" (also referred to as a "sample"). If we convert the sampling rate (up or down), the word length remains the same. (Converting sample rate and changing word length are two different processes.)

 

Word length (sometimes referred to as "bit depth") describes how many digital bits are used in each digital "word". So, 16-bit audio at 44.1 kHz sampling rate (the CD standard) will contain 44,100 "samples" (or "words") per second and each one is 16-bits long (e.g., 1011010111110001). If we padded the example in the previous sentence to 24-bits, it might look like this: 101101011111000100000000. Once we processed that 24-bit signal, the sample in the previous sentence might look like this: 101101011111000101011001.

 

Hope this helps.

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

 

Helped me understand better generally.

 

With appreciation for making what I tend to glaze over, understandable, relatively speaking. ;>}

 

Best,

Richard

Link to comment

Hi Barry,

 

thanks for taking the time to explain. I have in the meantime gained a better understanding of why 32 bit float has an advantage in terms of headroom, from you and also from here:

Even if 32 bit float is still 24 bit, as you said.

 

I have also contacted Audirvana, who kindly responded and confirmed that the Macbook Pro with Audirvana Integer mode is really capable of sending 32bit data to my DAC, because it bypasses the limitation of CoreAudio which uses 32bit float data, which is limited to 24bit precision.

 

This would mean that the limit here is the Hegel DAC, which, like most (nearly all?) DACs, does not accept 32 bit data. However insofar not a problem as, as stated before, sonically 32 bit float files sound clearly better. So will keep recording in this format!

 

I forgot to say that I recorded in 32/44.1, which keeps memory at still reasonable levels and facilitates a transfer to the 16bit iPod for mobile listening. I can’t hear frequencies above 15 khz any more, and my speakers don’t go linear into the 30-40 khz region, therefore agree that all the 96-192-384 hype is bollocks. From the tests I conducted so far am left with the impression that higher bit rates are much more important than higher sample rates.

 

Could it be that the current 24 bit or DSD formats haven’t yet reached the limits of what a humble human ear is capable of hearing?

 

Best,

 

Roberto

Link to comment

Hi Roberto,

 

Hi Barry,

 

thanks for taking the time to explain. I have in the meantime gained a better understanding of why 32 bit float has an advantage in terms of headroom, from you and also from here:

Even if 32 bit float is still 24 bit, as you said.

 

I have also contacted Audirvana, who kindly responded and confirmed that the Macbook Pro with Audirvana Integer mode is really capable of sending 32bit data to my DAC, because it bypasses the limitation of CoreAudio which uses 32bit float data, which is limited to 24bit precision.

 

This would mean that the limit here is the Hegel DAC, which, like most (nearly all?) DACs, does not accept 32 bit data. However insofar not a problem as, as stated before, sonically 32 bit float files sound clearly better. So will keep recording in this format!

 

I forgot to say that I recorded in 32/44.1, which keeps memory at still reasonable levels and facilitates a transfer to the 16bit iPod for mobile listening. I can’t hear frequencies above 15 khz any more, and my speakers don’t go linear into the 30-40 khz region, therefore agree that all the 96-192-384 hype is bollocks. From the tests I conducted so far am left with the impression that higher bit rates are much more important than higher sample rates.

 

Could it be that the current 24 bit or DSD formats haven’t yet reached the limits of what a humble human ear is capable of hearing?

 

Best,

 

Roberto

 

 

 

 

Either I'm misunderstanding something or you are being told something which is not true.

The idea of "32-bit" audio is nonsense. We are lucky when the hardware or software can produce 24-bit audio and get remotely close to 24-bits. (Much hardware and software in my experience, does not get close to 24, regardless of what the marketers would like us to believe.) Put another way, no one, of whom I'm aware, is making 32-bit recordings.

 

If you get 24-bit PCM, you are doing well. There are no 32-bit DACs because there is no 32-bit audio. Anyone that tells you different either knows something I am not aware of (always a possibility) or they are talking nonsense.

 

My own experience (which may not be the same as yours) is that properly done 24-bit at 192k can be wonderful -- with the best hardware and software, to the point where I have not yet been able discern my mic feed from the 24/192 playback. But it has also been my experience that most gear claiming these numbers does not achieve their potential.

 

32-bit processing can be useful and a lot of consumer software can achieve this. The best pro software will operate internally at 48, 64 or even 80-bits. But the audio, if we're lucky, is 24-bits. What feeds the DAC is 24-bits. If the DAC is well designed, it is working at 24-bits. In my opinion, talk of 32-bits to the DAC, when it isn't silly, is just make believe at best and cynical b.s. fed to consumers at worst.

 

Personally, talk of the "limits" of Core Audio also makes me chuckle. The best recordings I've ever heard -- the ones where I can't tell them from the direct mic feed -- were made using Core Audio.

 

I'm always open for something better but my experience has shown me that the "better" is the rare exception, while the *talk* (claiming better) is common.

 

The very best I've heard in 2013, by a long, long margin, is properly done 24/192 (and this is still a rarity). So when I hear claims of 32-bits or 768k, I assume the source of such claims is either completely unaware of the reality or would like to believe that the consumer is unaware.

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

Link to comment

Oh, no, the situation is even worse. Right now, there are just a few DAC chips that can reach true 22 bit analog resolution. Sure you can cram 24 or 32 bit in their interfaces, but those 'lower' bits will be just covered in noise and distortion until they reach the 20-22nd bit levels (counting from 'top').

In studio processing, sure, they can use a 64 bit path if they want, to cover all those manipulations and round-up errors. But this is digital domain, when you hit the ear, you still need and Digital-To-Analog module, because we hear analog signals only.

 

PS: For reference, 22 bit true resolution means a THD+N lower than -132dB, 21 bit means better than -126dB, 20 bit means better than -120dB. Compare that with your favorite DAC numbers.

Link to comment

Friends I'm just a normal audiophile with some grey matter left between my ears, am totally in your hands to learn more here. So Audirvana's response claiming 32 bit output is wrong? And when Hegel wrote to me saying they were not aware of any computers outputting 32 bit - so they are right? I tried to play the same '32 bit' files with Amarra (not the latest version), but there it did not work, I got only loud noise.

I checked again, believe i said something inaccurate before: my RME UCX allows 24 bit and 32 bit float, but not more.

Barry am impressed by your 80 bit recordings - are you recording 80 bit float? And can you hear a difference by going so high? Wow guys..

Link to comment

Audirvana claim 32 bit output is deceiving. Because it is a pure digital software it's can 'output' whatever they want. It doesn't mean that you will be able to use that effectively.

Like I said, in mixing/studio work makes sense to use a wider path to perform the desired (digital) processing.

But the final limitation is at the listener - DAC stage - that's where you don't have means to achieve more than 20-22 bit of real analog resolution.

Link to comment

PS: For reference, 22 bit true resolution means a THD+N lower than -132dB, 21 bit means better than -126dB, 20 bit means better than -120dB. Compare that with your favorite DAC numbers.

 

That is quite interesting statement, because even the top end DACs cannot usually do even -120dB?!

--

Krzysztof Maj

http://mkrzych.wordpress.com/

"Music is the highest form of art. It is also the most noble. It is human emotion, captured, crystallised, encased… and then passed on to others." - By Ken Ishiwata

Link to comment

Hi Roberto,

 

...Barry am impressed by your 80 bit recordings - are you recording 80 bit float? And can you hear a difference by going so high? Wow guys..

 

I'm sorry if I was not clear in my earlier post. (I *thought* I was.) What I said was that one of the main tools I use in my recording and mastering work has 80-bit (fixed point, not float) data paths. This refers to the precision with which it performs its internal math. (Yes, when processing a digital audio signal, I find this tool to provide quite clearly superior results. I would, however, attribute this to the whole design and not simply to that one aspect of it described by the word length it uses.) I believe I also said that when I'm listening via this tool, the signal it feeds to my DAC is 24-bit and the audio I hear is 24-bit.

 

Same when I record through this tool -- the file I save is 24-bit. What I've tried to communicate in my posts in this thread is that we are lucky if a software application or hardware device can actually deliver 24 clean bits. (I am *not* referring to the signal-to-noise specifications others in this thread have referred to.) With the best software and hardware, we can get 24-bit audio (if not necessarily the 144 dB signal-to-noise ratio theory suggests).

 

In one sentence: In my opinion, claims for audio (as opposed to internal processing) beyond 24-bits are a marketer's numbers game, designed to impress those who think a bigger number provides better sound, instead of just a bigger number (and sometimes, quite *inferior* sound).

 

Best regards,

Barry

http://www.soundkeeperrecordings.com

http://www.soundkeeperrecordings.wordpress.com

http://www.barrydiamentaudio.com

Link to comment
That is quite interesting statement, because even the top end DACs cannot usually do even -120dB?!

 

Exactly. What's funny is that some of the newer '32 bit' chips have lower performance than the 24 bit chips of the same manufacturer.

 

Well, that is a bummer!

--

Krzysztof Maj

http://mkrzych.wordpress.com/

"Music is the highest form of art. It is also the most noble. It is human emotion, captured, crystallised, encased… and then passed on to others." - By Ken Ishiwata

Link to comment

I don't own any 32 bits music track. On PCM the maximum is DXD (24/352.8) from 2L Recordings. I don't know if they were recorded on 24 or 32 bits, but their DLs are 24. I saw some esoteric (not the brand) CDs selling from 34/352.8 masters...? It could be possible but there is no way to proof this. Maybe it is a marketing strategy as Barry said, who knows?

 

In the case of Audirvana Plus, it's 64 bits player per se (as is iTunes now) matching Mountain Lion capability, but you can play it (as iTunes) on 32 bits, just to match older OS X systems. This has nothing to do with the music track bit output, but only to work better with the OS X used.

 

Regarding A+ outputting 32 bits is to match your DAC capabilities, for example to some ESS Sabre 32 bits DACs:

 

ESS Products - DAC

 

I guess this 32 bits DAC capability is to get their filter/algorithms and 8 channels possibility at their best.

 

In A+ to get an 32/xxx output depends on DAC USB interface. It must be integer/direct mode capable, then not all the ESS Sabre based DACs are 32/xxx capable.

 

I rather not state which one has the better SQ because this could be (as it is) user/system dependent. Also (the last but not the least) THD figures means nothing to me, but the SQ that please my ears. But it seems a lot of people are very impressed by stated numbers...

 

Roch

Link to comment

Hi Roch,

 

...In the case of Audirvana Plus, it's 64 bits player per se (as is iTunes now) matching Mountain Lion capability, but you can play it (as iTunes) on 32 bits, just to match older OS X systems. This has nothing to do with the music track bit output, but only to work better with the OS X used....

 

Yes, exactly! A common misunderstanding I've been seeing on audio fora is the confusion between a program that uses say, 64-bit addressing, to take maximum advantage of a modern operating system and 64-bit audio. Perhaps this too, is used to advantage by unscrupulous marketers, selling to innocent audio enthusiasts who have been trained to believe a bigger number *must* mean better results.

 

Like the marketers of "32-bit DACs", these folks are *not* talking about what you and I can listen to in the real world.

(DXD *may* be an exception. Maybe. I don't know. A lot of 192k DACs reveal just how difficult it is to design a product that can truly show the potential of this rate. That makes me more than a little bit skeptical of claims for higher rates.)

 

The iTunes application, like many Apple apps, has been updated to make use of 64-bit addressing. Again, this has to do with how the application interacts with the operating system and is completely unrelated to digital audio wordlength of files we listen to. In this regard, iTunes can play 24-bit files.

 

This should not be confused with an audio application like Reaper, which does its internal processing of the audio at 64-bits. Here again, lest anyone start to think they will hear "64-bit audio", the signal fed to the DAC -- the signal we listen to -- is 24-bits. It has merely been processed with higher precision internal math, IF any processing has been applied by the Reaper user.

 

Hopefully, folks will begin to understand the BIG difference between an app that utilized 64-bit addressing (becoming common), an app that utilizes 64-bit data paths for its internal math (several pro applications do) and 64-bit audio (which in the here and now, where in many quarters, 24-bit is pushing it, is a fable).

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

Link to comment

Well all of this if highly interesting, problem is how many people will actually understand it? What I see when clicking for info in iTunes is '32 bit 44.1 khz'. So mister hifi music lover believes that the file he listens to contains 32 bit audio data.

 

And then, looking at the Audio Midi Setup where the DAC is identified as “32 bit”, mister hifi music lover comes to the same conclusion.

 

And finally, according to the Audirvana manual, at the bottom left of the screen it shows the playing track file type, native bit depth and sample rate, indicating “AIFF-C 32/44.1 khz”, and on the right it shows the DAC's current bit depth and sample rate, indicating “32/44.1 khz”. So mister hifi music lover reaches the same conclusion for the third time, and is now really reassured that what he has and listens to is 32 bit sound.

 

Until now.

 

I am grateful for Barry and the others in this forum for sharing their thoughts and experiences so we can dig a - single - bit deeper, but should there not be stricter rules for hardware and software companies presenting their products, bit lengths and sample rates?

Link to comment

Hi Roberto,

 

...should there not be stricter rules for hardware and software companies presenting their products, bit lengths and sample rates?

 

If a display, such as the Audio/MIDI Setup control, is sent "32 bit", that is what it will display. It is not reading the audio.

The DAC manufacturer talks about "MIDI settings" on their site (and calls out "24-bit") but the DAC is *not* a MIDI device. (Perhaps this is a language issue. I would expect a high end manufacturer to avoid problems like this.)

The manual for the DAC also claims "32 bit" for the "DAC IC/Digital Filter" but note under "Digital Inputs" it says "Up to 24 bit".

 

This is like the CD label that claims "100k mastering". I am guessing they use 96k sampling somewhere in the process and are simply rounding up the number. (Or perhaps they use non-standard digital hardware/software?) Yet the CD they sell is 44.1k, no more. It would be like my calling Soundkeeper CDs "80-bit/200k mastering" because I have 80-bit tools and work at 192k. That, in my view, would be just as silly because the final CD is 16-bit (no more) and 44.1k (no more).

 

The stricter rules you ask about would, in my opinion, simply be called "honesty". ;-}

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

Link to comment

CoreAudio format is 32bit float. And this is actually 24bit precision (32bit float = 24bit mantissa (integer) + 8 bit exponent).

This format is the only one that is exposed to audio applications on OS X (iTunes, Logic, all of them).

Except for Audirvana Plus that is bypassing this CoreAudio Layer to directly send the audio data in the DAC native format.

This is nowadays usually 24bit, or 24bit embedded in 32bit chunks (displayed as 24bit) for ease of manipulation by the CPU.

 

Some audio interfaces tell the Mac to send their data in 32bit. Yes, 32bit. If Audirvana Plus was sending the data in another format you would get noise as this 32bit is the format the DAC computer interface excepts the data to be.

Now this doesn't mean a) the DAC chip will actually get 32bit data (it is truncated inside the DAC to 24bit before reaching the converter itself) b) if it does, as Barry and others mentioned this does not guarantee better quality.

 

This audio interface format also accounts for why most of the time, the 16bit data needs to be padded to send 24bit to the DAC interface that recognizes only 24bit.

 

Now regarding 64bit. This is the addressing and instruction set of the CPU of modern computers for better performance of course. But, for Audirvana Plus, it is also the internal processing, as in the studio softwares, to avoid as possible any rounding error when performing filters like sample rate conversion. The audio signal is of course converted to the audio interface format at the end of the processing chain (e.g. 24bit).

AFAIK iTunes uses only 32bit processing precision for its internal processing.

 

Damien

MBP 15"/Mac Mini, Audirvana Plus, Audioquest Diamond USB, AMR DP-777, exD DSD DAC (for DSD), Pioneer N-70AE, Audioquest Niagara balanced/Viard Audio Design Silver HD, Accuphase E-560, Cabasse Sumatra MT420

Link to comment
CoreAudio format is 32bit float. And this is actually 24bit precision (32bit float = 24bit mantissa (integer) + 8 bit exponent).

This format is the only one that is exposed to audio applications on OS X (iTunes, Logic, all of them).

Except for Audirvana Plus that is bypassing this CoreAudio Layer to directly send the audio data in the DAC native format.

This is nowadays usually 24bit, or 24bit embedded in 32bit chunks (displayed as 24bit) for ease of manipulation by the CPU.

 

Some audio interfaces tell the Mac to send their data in 32bit. Yes, 32bit. If Audirvana Plus was sending the data in another format you would get noise as this 32bit is the format the DAC computer interface excepts the data to be.

Now this doesn't mean a) the DAC chip will actually get 32bit data (it is truncated inside the DAC to 24bit before reaching the converter itself) b) if it does, as Barry and others mentioned this does not guarantee better quality.

 

This audio interface format also accounts for why most of the time, the 16bit data needs to be padded to send 24bit to the DAC interface that recognizes only 24bit.

 

Now regarding 64bit. This is the addressing and instruction set of the CPU of modern computers for better performance of course. But, for Audirvana Plus, it is also the internal processing, as in the studio softwares, to avoid as possible any rounding error when performing filters like sample rate conversion. The audio signal is of course converted to the audio interface format at the end of the processing chain (e.g. 24bit).

AFAIK iTunes uses only 32bit processing precision for its internal processing.

 

Damien

 

Hi Damien,

 

Believe or not last night I have a dream about you explaining this...

 

Thanks again for your great job in A+,

 

Roch

Link to comment
  • 6 months later...

Hello,

 

Is it that the signal occupies 24-bits, and in the case of some DACs with higher bit processing that the extra bits are reserved for digital volume control without bit truncation?

 

I.e. From the NAD M51, according to NAD, "The extreme headroom afforded by the 35-bit architecture allows for a DSP-based volume control that does not reduce resolution. Even with 24-bit high definition signals, the output can be attenuated by 66dB (very, very quiet) before bit truncation begins."

 

FWIW, I've owned three DAC's:

NAD M51

AMR DP-777

Hegel HD25

 

And the Hegel stands head & shoulders above the other two. The hegel has very well-behaved tonality, with slight feel-good warm bump, and I'm guessing around 150Hz. The other two had some tonality unevenness; the M51 sounded dry or mechanical, and the DP-777 had some tonality unevenness, sounding thin. The Hegel HD25 is a keeper!

 

 

Link to comment

If you're doing operations with 32 bit "words," then yes, just as with Barry's 80 bits, you can do DSP such as digital volume control with less, possibly no audible, signal deterioration.

 

But you *never* hear 32 bit or even 24 bit audio. As Sonic says, the very best equipment has residual thermal noise somewhere between 20-22 bits. Below that all that's being played back, too low for anyone to hear, is entropy.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
If you're doing operations with 32 bit "words," then yes, just as with Barry's 80 bits, you can do DSP such as digital volume control with less, possibly no audible, signal deterioration.

 

But you *never* hear 32 bit or even 24 bit audio. As Sonic says, the very best equipment has residual thermal noise somewhere between 20-22 bits. Below that all that's being played back, too low for anyone to hear, is entropy.

 

Hi Jud,

 

These claims remind me of those who are talking 384k and 768k sampling. (Besides, 512-bit/1536k is much better. ;-})

 

If one considers how much trouble so many designers appear to be having in making 192k DACs that can actually deliver the potential of that rate, claims of higher rates, when aren't sad, are hilarious. So many of the DACs out there today get "brittle" at 4x rates (like 176.4k and 192k). And so many sound pretty cheesy in the low order bits of a 24-bit signal. When those two things become more commonly achieved with success, I won't see claims of the higher numbers as empty at best and cynical at worst.

 

From the home of 80-bit/200k recordings. ;-} (That's a joke folks.)

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

Link to comment

 

Hi Jud,

 

These claims remind me of those who are talking "768k sampling". (Besides, 1536k is much better. ;-})

 

If one considers how much trouble so many designers appear to be having in making 192k DACs that can actually deliver the potential of that rate, claims of higher rates, when aren't sad, are hilarious. So many of the DACs out there today get "brittle" at 4x rates (like 176.4k and 192k). And so many sound pretty cheesy in the low order bits of a 24-bit signal. When those two things become more commonly achieved with success, I won't see claims of the higher numbers as empty at best and cynical at worst.

 

From the home of 80-bit/200k recordings. ;-} (That's a joke folks.)

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

 

Hi Barry,

 

Is it correct to understand you as talking about the ADC rather than the DAC end of the process, since 8x oversampling and delta-sigma modulation are pretty standard for the latter?

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...