Hi-Rez - Another Myth Exploded!

[gregorio]

The dry signal isn't clipped.  It's a dirac pulse.
Also, this VST is from Lexicon lol.  I paid about 1.5k for it.
 
Oh!  I see what happened!  The problem lies in what the oversampling algorithm I used did to the pulse!
 
I fixed it.  They now have the same pulse!
 
http://www.pelpix.info/native.flac
http://www.pelpix.info/oversampled.flac
 
They now sound exactly the same except for the high-end roll-off!

Now they seem much more similar. There is obviously less HF in the native version. I can't tell if the HF is noise, distortion or actual generated reflections, I would have to listen in my studio to get a more accurate idea of what's going on. Is the Lex oversampling at 96kS/s or 192kS/s?

G

 

[PelPix]

Quote:

Now they seem much more similar. There is obviously less HF in the native version. I can't tell if the HF is noise, distortion or actual generated reflections, I would have to listen in my studio to get a more accurate idea of what's going on. Is the Lex oversampling at 96kS/s or 192kS/s?

G

 
176.4kS/s.  Even people that support oversampling know not to oversample to non-multiples 

 

[gregorio]

176.4kS/s.  Even people that support oversampling know not to oversample to non-multiples 

I should have said upsampling rather than oversampling. Your statement here is much less true today than it was 10 years ago. Technology has improved and decent ASRC is far less detectable than it used to be (depending on the quality of your ASRC processor). Myself and many others in the professional recording industry have discovered that quite a few plugins sound better when operated at 96kS/s, this is simply due to the fact that some manufacturers have concentrated their programming efforts at this sample rate at the expense of other sample rates. This also explains why there is sometimes an audible difference between running a plugin at 96kS/s compared to 88.2kS/s. As I've said before, it's all about trade-offs. Do the benefits of running plugins at 96kS/s outweigh the disadvantages of ASRC? The answer is often yes. If we ask the same question of 192kS/s the answer is far more straight forward. I know of no plugins which sound better at 192kS/s than at 96kS/s. This makes sense because it would be nonsensical for a plugin manufacturer to concentrate their efforts on 192kS/s for two simple reasons: 1. 96kS/s is far more commonly used than 192kS/s and 2. The number and complexity of executable instructions is compromised by the high speed requirements of 192kS/s. On the balance of trade-offs, it usually makes sense to record and mix at 24/96kS/s and then (for the mastering engineer) to SRC/dither to 16/44.1 for distribution.

It is extremely unlikely (read impossible) therefore that on the balance of trade-offs a mix would benefit from 192kS/s processing compared to mixing at 96kS/s.

Just a note to anyone else reading. We are discussing the performance of digital signal processors employed during the mixing and mastering phases. Although the issues being discussed are directly related to some of the same principles as in the OP, they are not directly related to consumer playback. While there is good reason to record at 24/96 and to mix in a 48bit 96kS/s environment, I maintain (backed by various studies) that when competently converted to 16/44.1 the difference is inaudible. With the caveat that your DAC has well implemented 44/1kS/s filters. And on the off chance I've not been clear, my position (based on the scientific evidence) is that 176.4kS/s and higher sample rates are always inferior for any use (recording, mixing, processing, mastering and playback).

G

 

[judmarc]

Looking back over what I wrote it seems a touch confrontational
 
Not at all.  Hope you'll take my responses in the same constructive spirit.
 
Whatever ideal assumptions contained therein are nothing but mere math, which is ubiquitous in reality. More to the point, Shannon/Nyquist isn't meant to be an ideological force, it is more of a naturally observed limit, much like we can calculate the maximum speed of a photon through a vacuum. The practical limitations of the theorem are why we need more than the 10% headroom afforded by 44.1khz.
 
Let me be a little more precise about what I meant.  From our friend Wikipedia on Nyquist-Shannon:
 
The theorem assumes an idealization of any real-world situation, as it only applies to signals that are sampled for infinite time; any time-limited x(t) cannot be perfectly bandlimited. Perfect reconstruction is mathematically possible for the idealized model but only an approximation for real-world signals and sampling techniques, albeit in practice often a very good one.
 
So - headroom because we don't have infinite time, and headroom because of the real-world limitations of filters ("brickwall" bandlimiting produces audible artifacts if done near 44.1kHz).  But notice that we are no longer in the realm where we can settle back and say "Good!  A little headroom and now we can do perfect reconstruction under Nyquist-Shannon."  We can only make a better (in practice very good) approximation, and use filtering that won't harm the audio signal as much.  So we can't say with mathematical certainty we've reached perfection at 96kHz and no more improvement is possible.  To use your analogy, we've made a better (but not a perfect) vacuum, so the speeds we measure for photons are a better approximation of the theoretical limit but not mathematically equal to it.
 
Can we make an audibly better approximation at 176.4 or 192 than at 88.2 or 96?  Some people think so, some don't; I don't know for sure.  But I haven't seen anything that tells me we must absolutely foreclose the possibility.
 
This isn't esoteric knowledge and anyone with a high tolerance for dry reading can go to the library and check out a textbook on this stuff that says the same thing.
 
I'm quite curious about the subject and would much appreciate recommendations/citations if you have them.
 
I see a fair bit of cynicism in your comments which appears to hide a lack of familiarity with the science
 
That would be an incorrect conclusion.    Rather, what you find "cynical," I would characterize as paying attention to Santayana.  We don't want to doom ourselves to repeat the history of "perfect sound forever" by ignoring what happened then.  What happened?  It seems to me we thought our "very good approximation" of Nyquist-Shannon was so close that for practical audio purposes we'd achieved perfect reconstruction, or as near as it was possible to get.  We thought so strongly enough that many folks were willing at first to ignore audible evidence.  Later of course the audible became measurable (e.g., jitter, "brickwall" filtering, and resultant distortion in the audio band).
 
Now we certainly take more care with jitter and filtering, and we've given ourselves a lot more headroom.  If a model should be as simple as possible but no more, is it now time to say we've got everything important in our model at 96kHz, and any difference between that and 192kHz cannot be audible and therefore cannot be important enough to include?  As I said above, some people think so, some don't, and I don't know for sure.
 

 

[jcx]

 
Quote:

Originally Posted by judmarc /img/forum/go_quote.gif
 
...I see a fair bit of cynicism in your comments which appears to hide a lack of familiarity with the science
 
That would be an incorrect conclusion.    Rather, what you find "cynical," I would characterize as paying attention to Santayana.  We don't want to doom ourselves to repeat the history of "perfect sound forever" by ignoring what happened then.  What happened?  It seems to me we thought our "very good approximation" of Nyquist-Shannon was so close that for practical audio purposes we'd achieved perfect reconstruction, or as near as it was possible to get.  We thought so strongly enough that many folks were willing at first to ignore audible evidence.  Later of course the audible became measurable (e.g., jitter, "brickwall" filtering, and resultant distortion in the audio band).
 

not exactly correct, there are accounts of Sony/Philip's gaming the standards for their own corporate advantage - the engineers at the time knew that compromises that could be problematic were being made to fit the existing prototype CD data capabilities, some "requirements" were unfounded, completely arbitrary, decided in the usual human, messy political process
 
you are confusing the later Marketing of CD audio to the reviewers, buying public with the audio engineering knowledge of the technical reality - Marketing is where your cynicism should be directed
 
and should be pointing today to the foolishness of Beatles 24 bit 44.1 releases being touted as "hi res" and a necessary, wallet emptying advance
 

 

[judmarc]

Judmarc I've been doing "hi-rez" digital recording professionally for nearly 20 years, in fact many years before it started to be called hi-rez. I have done work at a number of the world's top recording studios, with some of the very finest equipment in the world. I am a recording engineer and producer, I do not have PhDs in maths, Electronic Engineering and Physics so there's a limit to my understanding of digital audio, even if I have picked up a fair bit along the way. But, you seem to think that everything I'm saying is just quoting Dan Lavry. I'm quoting Lavry because his paper was the first to tackle this issue and it is widely accepted by the scientific and recording communities. Even some of the other manufacturers (Prism, Benchmark and Apogee) have publicly endorsed Lavry's paper, and I believe Weiss has too, certainly their top of the line pro converter does not support sample rates >96kS/s. Lavry, Prism and Weiss represent the top professional converter manufacturers in the world.
 
Regarding not relying completely on Lavry, fair enough.
 
Regarding Weiss, their currently offered professional ADC and DAC both provide 176.4 and 192kHz capabilities, as shown on their web site.
 
Nearly 20 years, eh?  Maybe not quite the beginning (Ry Cooder's Bop Til You Drop, the first major label digitally recorded album, came out in 1979 - great record, by the way), but a long time.  
 
So maybe a couple of references to Mr. Johnson's work will strike a familiar note:
 
- With Michael "Pflash" Pflaumer, invented and patented HDCD.
 
- With Pflaumer, started Pacific Microsonics, and developed the Model One and Model Two ; more detail available at http://www.goodwinshighend.com/manufacturers/pacific_microsonics/pacific_microsonics_model_two.htm .  Perhaps you've seen or used the Model One or Model Two in the various studios in which you've worked?
 
- With Pflaumer, developed the Berkeley Audio Design Alpha DAC.
 
This isn't meant as some sort of battle of reputations.  It's just meant to say that there are some *very* smart folks who know their way around digital audio design, recording and playback who think there's an audible advantage to resolutions beyond 96kHz, just as there are such folks who think 96kHz is as good as it gets for all practical audio purposes.
 
So as of right now, I'm withholding judgment.

 

[gregorio]

t's just meant to say that there are some *very* smart folks who know their way around digital audio design, recording and playback who think there's an audible advantage to resolutions beyond 96kHz, just as there are such folks who think 96kHz is as good as it gets for all practical audio purposes.

Yes, my mistake on the Weiss. In actual fact, Lavry and Benchmark both produce 192kS/s products too. Isn't this hypocritical? The problem is, there are only a handful of DA chip manufacturers, so pretty much, without some serious engineering, as a DAC manufacturer you are stuck with the sample rates provided by the DA chip manufacturers. Unfortunately that means 192kS/s but as Benchmark says: "All of Benchmark’s A/D converters and D/A converters support sample rates up to 192kHz. However, we strongly recommend 96kHz for optimum performance."

I have heard of and seen the Pacific Microsonics HDCD, and I know of Berkeley Audio.

Let's, for the sake of argument say that hypothetically there are no physics or processing limitation problems with sample rates greater than 96kS/s. That leaves us with a 192kS/s sample rate which performs just as well as 96kS/s with the added benefit of being able to record frequencies between 48kHz and 96kHz. My dog can't hear 48kHz and even some bats can't hear as high as 96kHz. Between 48kHz and 96kHz we are looking at virtually zero energy produced by musical instruments, no standard studio mics which can record that high and virtually no cans or speakers (and probably not many amps) which can accurately reproduce 96kHz, in addition of course to the fact that 96kHz is nearly 5 times beyond the limit of human hearing. So even if there weren't any problems with 192kS/s it would still be absolutely pointless. I would like you to ask your Mr. Johnson how he EQ's and mixes all these frequencies which he can't even hear?

G

 

[khaos974]

To be fair, I doubt processing files with a 192 kS/s with plugin is an issue if one doesn't need real time, after all CPU power isn't bound by the same limits as physical components, we stille very much enjoy a doubling of CPU power every 18 months and will probably continue with Moore's law for the next ten years (at least).

 

[gregorio]

So - headroom because we don't have infinite time, and headroom because of the real-world limitations of filters ("brickwall" bandlimiting produces audible artifacts if done near 44.1kHz).  But notice that we are no longer in the realm where we can settle back and say "Good!  A little headroom and now we can do perfect reconstruction under Nyquist-Shannon."  We can only make a better (in practice very good) approximation, and use filtering that won't harm the audio signal as much.  So we can't say with mathematical certainty we've reached perfection at 96kHz and no more improvement is possible.  To use your analogy, we've made a better (but not a perfect) vacuum, so the speeds we measure for photons are a better approximation of the theoretical limit but not mathematically equal to it.
 
Can we make an audibly better approximation at 176.4 or 192 than at 88.2 or 96?  Some people think so, some don't; I don't know for sure.  But I haven't seen anything that tells me we must absolutely foreclose the possibility.


We don't want to doom ourselves to repeat the history of "perfect sound forever" by ignoring what happened then. What happened? It seems to me we thought our "very good approximation" of Nyquist-Shannon was so close that for practical audio purposes we'd achieved perfect reconstruction, or as near as it was possible to get. We thought so strongly enough that many folks were willing at first to ignore audible evidence. Later of course the audible became measurable (e.g., jitter, "brickwall" filtering, and resultant distortion in the audio band).

I missed this post before and you make some good points, although maybe a little inaccurate in some of your statements.

I agree with jcx. It is not accurate to say that the science and the pro audio community thought CD was perfect and then found ways to measure the imperfections. Accurate scopes existed in the early 80's, the weakness of digital audio implementation was well known (both measured and audible) and the CD Redbook standard was hotly contested at the time. Despite the known weaknesses, CD was marketed as perfect. Again, as jcx states the whole thing was commercial/political, with the manufacturers winning out. For this reason, the perfect digital audio playback format does not even exist! The perfect balance of speed vs accuracy for the consumer would be about 16bit 60kS/s. You are making a mistake which is rife in the audio world, confusing the marketing of facts with the actual facts. It was for this reason more than any other than I started this thread!

Have we now reached perfection according to Nyquist/Shannon? Depends on how you define perfection; mathematically no we haven't but perfection to the point of substantially exceeding the ability of the ear to detect, yes. DACs have existed for quite a few years which measure a linear response to a point far beyond audibility. Jitter, such a problem according to much marketing, is a non-issue with any decently designed DAC. To the point that highly expensive, high precision equipment is needed to be able to measure it (at the point of the AD or DA chip) and orders of magnitude below what is audible. Bit depth (dynamic range) has already exceeded not only the ability of the human ear but even the laws of physics. Despite DACs being marketed as 24bit no DAC on the planet can actually resolve more than about 21 bits. A dozen or so years ago 18 bit was about the maximum technology was capable of resolving but at 21bit the limit is no longer technology but the laws of physics, so this is never going to change. That doesn't seem to have stopped the latest fad of marketing 32bit DACs. If 32bit dynamic range was actually possible it would kill you instantly! The same is true of sample rate, although fortunately you can't be killed (or injured) by higher sample rates.

By far the biggest problem facing the digital audio world today is actually the consumer and the manufacturers' ideas of what the consumer wants!!!! Rather than stopping at the point of digital audio perfection we just keep on going, apparently oblivious to the fact that we've missed the stop and are now travelling further and further away from where we should have got off. Audibly perfect DACs have existed for some time but the development of technology to make that level of perfection cheap and ubiquitous is a dead end road for marketing departments.

So am I saying we have reached the end of the road for better quality audio? Not at all! On the technology front there is still improvement to be made; with digital the cheap availability of audibly perfect DACs but even more can be done on the analogue side, particularly with transducers and acoustics for example. But dwarfing just about all other considerations put together, the biggest improvement by far which can be made is with the consumer. There are several areas for the consumer which need improvement if audio quality is to be increased:

1. The realisation that the experience of a live musical performance is not dependent on sound waves alone and therefore will never be captured or reproduced by digital audio technology.
2.. The ability of the consumer to appreciate, demand and pay for higher quality music products.
3. A demand for higher quality equipment at reasonable prices rather than for equipment with bigger numbers.

Point 2 is particularly important. The consumer has demonstrated that in general it is more interested in low or no cost, than it is in quality. The demand for MP3s at prices of $1 a song and for those MP3s to be as loud as possible (or indeed most of the time, IMO, louder than is possible). Obviously I'm talking about the consumer in general rather than the relatively niche audiophile market. Audiophiles though are more responsible for points 1 and 3 above. It seems to me that many audiophiles (maybe even a majority) do not want perfectly linear systems, they want distortion. Presumably because of past experience of other sound systems (consumer systems and/or live sound re-enforcement systems) many audiophiles do not seem to measure the quality of a sound system by it's linearity of reproduction. Preferring instead warmth, HF distortion, noise, etc. The love of vinyl, tubes, coloured cans and speakers, filterless DACs, etc., is proof of this. I've seen descriptions from audiophiles about DACs (or other equipment) such as too analytical, too detailed or too clinical. A DAC can't produce more detail than exists in the recording and cannot therefore be too analytical or clinical. The audiophile is basically saying they don't like the sound of linear reproduction. Many audiophiles seem to be highly influenced in their appreciation of what constitutes quality by marketing tactics such as pseudo science, shills and reviewers with advertising revenue at stake. Therefore, I'm sure some (if not many) will find what they are looking for in 24/192, the upcoming 32/384 and presumably in 64/768 (as soon as someone finds a way to make it available for marketing).

The relatively few knowledgeable audio professionals and audiophiles just do not represent a big enough consumer marketplace to make linear playback and high quality recordings anything other than a minority of an already niche market. Until real quality (rather than a marketing led concept of quality) becomes profitable we are going to see a continued decline in the quality of recorded music and movement further and further away from the Nyquist-Shannon ideal of digital audio recording and reproduction. Virtually none of the world's top commercial recording studios make an operating profit. If the current trends continue, it won't be so long before the music industry looses the knowledge, skill and facilities required to make a high quality music product, maybe forever or at least until real quality becomes fashionable (profitable) again.

G

 

[judmarc]

If 32bit dynamic range was actually possible it would kill you instantly! 
 
I love this image - kind of like the guy in that old Maxell ad sitting in a leather chair with his hair blowing back, only in this case he's blown right out the back wall! 
 
Actually, I think the supposed point of more bits of dynamic range used to be to allow the noise floor to get really, really low rather than the music to get really, really loud, but I agree with you that the point of practical utility has passed somewhere between 16 and 24 bits.  I'm thinking I may recall hearing that another reason for 32 bits was allowing finer software volume adjustment, but I have no idea at all whether that holds any water.
 
Regarding whether we yet have "audibly perfect" DACs, I haven't heard enough modern DACs myself yet to make a personal judgment on that, but my inclination would be to disagree.  If two DACs are "audibly perfect" they ought to sound identical, and I wonder whether we can find two that do?  My guess is we wouldn't be able to - but that's just a guess, not based on much listening, at least not yet, though I hope to find enough leisure time in my future to be able to listen to lots of very good (maybe even "audibly perfect"?) DACs.
 
I've seen descriptions from audiophiles about DACs (or other equipment) such as too analytical, too detailed or too clinical. A DAC can't produce more detail than exists in the recording and cannot therefore be too analytical or clinical. The audiophile is basically saying they don't like the sound of linear reproduction. 
 
We've been having a rather enjoyable discussion of this in a thread over at Computer Audiophile.  I'll say here the same thing I said there: If you don't like music from your "detailed, resolving, accurate, neutral" equipment, then your equipment is actually none of these things.  Real musical performances, in a concert hall or recorded in a studio and put together by a producer, are often engaging (or, depending on your taste, could be offputting), but above all are *different* from each other.  Different tracks on an album, let alone different albums by different artists, have very different sounds.  If you're constantly hearing lots of detail and resolution, even if the track was recorded with a lush, warm "feel," then your equipment isn't giving you what's on the recording and is by definition not accurate or neutral.  What such a piece of equipment is, is a "one-trick pony," and after owning it for some period of time, no wonder you're tired of its one trick!  The same with any component that renders everything as warm and lush.  I've just got a new CD from Gillian Welch, which is very well recorded, and most of the songs are produced to sound as dry as the Texas hill country.  Very little reverb on her voice, it just comes out of the speakers and gets right in your face.  Any piece of equipment that would make her sound like Barbara Streisand is doing something very wrong.  I'm sure many people could be momentarily infatuated with such equipment, but over time the constant sameness has just got to be fatiguing.  Ultimately, the most (truly) accurate piece of equipment is the one you're going to get the most enjoyment from in the long run.
 
I do have concerns about where the music and music reproduction industries are headed, same as you - the "noise wars," niche marketing to snobs as if audio equipment were fine wine - yep.  Whether higher resolutions than 96kHz are examples of the latter or can actually sound better, I'm looking forward to trying to figure out, starting when my Bifrost arrives.  (There are some very nice Linn classical pieces available in both 192kHz and 96kHz that won't cost too much to run my little experiment.)  Maybe I'll even settle back with a fine wine to do it (or as close as $12 a bottle at the local winery will get me).
 
Meanwhile, whenever anyone mentions Keith O. Johnson, just picture Bat Boy: http://en.wikipedia.org/wiki/Bat_Boy_(character) 

 

[gregorio]

Actually, I think the supposed point of more bits of dynamic range used to be to allow the noise floor to get really, really low rather than the music to get really, really loud, but I agree with you that the point of practical utility has passed somewhere between 16 and 24 bits.  I'm thinking I may recall hearing that another reason for 32 bits was allowing finer software volume adjustment, but I have no idea at all whether that holds any water.
 
Regarding whether we yet have "audibly perfect" DACs, I haven't heard enough modern DACs myself yet to make a personal judgment on that, but my inclination would be to disagree.  If two DACs are "audibly perfect" they ought to sound identical, and I wonder whether we can find two that do?  My guess is we wouldn't be able to - but that's just a guess, not based on much listening, at least not yet, though I hope to find enough leisure time in my future to be able to listen to lots of very good (maybe even "audibly perfect"?) DACs.
 

These days with noise shaped dither, the limiting factor is not the 16bit digital noise floor (-120dB approx) it's the noise floor of the mics, recording studio and speakers. There's an advantage to using 24bit when doing digital volume control but nothing really to be gained with 32bit.

Earlier this year, can't remember which trade magazine, Sound on Sound or Audio Media probably, they did a shoot out of modern ADC/DACs. DigiDesign (new and old 192s) Prism and Apogee converters with a bunch of top English engineers. The difference was tiny, I think they only just managed a significant result in DBT and these were some of the best ears, in a world class studio. I've not done any testing for a few years but certainly in my experience of pro converters, the differences have been getting smaller for the last 12 years, to the point that the differences are now vanishingly small. I don't think that now, even with my experience and decent studio, that I would be able to hear much, if any, difference between converters. The Lavry Gold blew me away when I heard it 10 years ago and is still my benchmark for the best 16/44 converter I've ever heard. I heard one again about two years ago and was still impressed but more from the fact it's stood the test of time, not sure I would be able to DBT it today.

G

 

[judmarc]

I think they only just managed a significant result in DBT 
 
Something I wonder about from time to time regarding double-blind testing, since hearing is a psychoacoustic phenomenon, is what questions are being asked, and what questions should be asked.  More specifically, are folks being asked questions like "Is this one the same or different from the last one?", or questions more like "Which of these do you like better, or do you feel there's no difference?"  The first sort of question I can see producing some pressure to get the answer right, perhaps not the best frame of mind in which to evaluate music.  I wonder if the second type of question would produce results that differed at all significantly (in either direction - being able to discriminate or not being able to) from the results produced by the first.
 
An interesting experiment for some grad student somewhere to perform, mebbe (and for others to try to replicate). 

 

[Willakan]

As I understand it, most DBTs simply request that the listener reliably distinguishes one piece of equipment from another - the question is simply "Which is which?" or "Which do you prefer?".
Indeed, in some tests people profess to still hear "impossible" differences - right up until their scores are added.

 

[Anaxilus]

Quote:

So am I saying we have reached the end of the road for better quality audio? Not at all! On the technology front there is still improvement to be made; with digital the cheap availability of audibly perfect DACs but even more can be done on the analogue side, particularly with transducers and acoustics for example. But dwarfing just about all other considerations put together, the biggest improvement by far which can be made is with the consumer. There are several areas for the consumer which need improvement if audio quality is to be increased:

1. The realisation that the experience of a live musical performance is not dependent on sound waves alone and therefore will never be captured or reproduced by digital audio technology.
2.. The ability of the consumer to appreciate, demand and pay for higher quality music products.
3. A demand for higher quality equipment at reasonable prices rather than for equipment with bigger numbers.

Great points to consider and I agree completely.  I've spent a good portion of my audiophile time obsessing w/ point #1.  I used to play the Piano and Violin and attend many live acoustic performances as well as amplified rock events too.  From day one I've felt in order to get that truly live sound some coloration is often needed if that is your goal.  Pure 'neutrality' or recording faithfulness won't do that and any recordings that even get you close are so few and far between your content selection becomes prohibitive.  I often link an article published in Stereophile that compared two electrostat headphones where they asked which was more accurate.  The recording was of a live Piano piece done in studio right then and there.  It was determined one stat sounded more accurate to the recording and the other more to the actual performance in the studio.  So IMO people need to understand your point and manage their expectations and understand the difference between a recording and a performance.  The balance needed in your gear to have the best synergy under the most possible conditions is not an easy thing to find which is why many like myself end up w/ more than one phone, source or amp.  If people can get away from the need to have simple, easy answers and open themselves up to the more insecure position of finding 'their sound' then understanding would improve and better dialogue and progress would be made. 
 
Quote:

 
Regarding whether we yet have "audibly perfect" DACs, I haven't heard enough modern DACs myself yet to make a personal judgment on that, but my inclination would be to disagree.  If two DACs are "audibly perfect" they ought to sound identical, and I wonder whether we can find two that do?  My guess is we wouldn't be able to - but that's just a guess, not based on much listening, at least not yet, though I hope to find enough leisure time in my future to be able to listen to lots of very good (maybe even "audibly perfect"?) DACs.
 
I've seen descriptions from audiophiles about DACs (or other equipment) such as too analytical, too detailed or too clinical. A DAC can't produce more detail than exists in the recording and cannot therefore be too analytical or clinical. The audiophile is basically saying they don't like the sound of linear reproduction. 
 
We've been having a rather enjoyable discussion of this in a thread over at Computer Audiophile.  I'll say here the same thing I said there: If you don't like music from your "detailed, resolving, accurate, neutral" equipment, then your equipment is actually none of these things.  Real musical performances, in a concert hall or recorded in a studio and put together by a producer, are often engaging (or, depending on your taste, could be offputting), but above all are *different* from each other.  Different tracks on an album, let alone different albums by different artists, have very different sounds.  If you're constantly hearing lots of detail and resolution, even if the track was recorded with a lush, warm "feel," then your equipment isn't giving you what's on the recording and is by definition not accurate or neutral.  

Agreed.  The analytical comments are not because something is too resolving but quite the contrary.  If something doesn't sound right it is because something is missing or unaccounted for.  It is not always about hype, placebo or whatever else.  Some things just sound wrong.  To my ears.  

 
Quote:

These days with noise shaped dither, the limiting factor is not the 16bit digital noise floor (-120dB approx) it's the noise floor of the mics, recording studio and speakers. There's an advantage to using 24bit when doing digital volume control but nothing really to be gained with 32bit.

This was always my sneaking suspicion and everything I tended to come across seemed to verify it.  There are other bits and pieces lost along the way, no pun intended, but those are the big ones.   
 

 

[goodvibes]

You can't just take dynamic range specs etc and not look at the entire picture. There's always going to be some discussion about what is audible and not but at 16/44, the amount of distorion or non linearity near the noise floor is quite high regardless of dither and effectively limits things. 16/44 has always sounded a bit dead or lacked that ease or sustain until masked by the next louder sound type of character I get from good analog or 24/96. I don't get more music going beyond 24/96. I have always found up or downsampling in exact multiples to be better and I use after market dither in Wavelab to do it.