State-of-the-Art Digital

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
It would, if they'd not taken the decision to effectively split the thing up into several boxes. That's what the combination of Vivaldi DAC, Upsampler and Clock effectively is: a DAC in three boxes, with isolated PSUs etc. The Bartok and Rossini, or even the Vivaldi One which includes an SACD transport, are one-box units that include upsampling capabilities. One may question the philosophy of splitting up the top model in four units (if one includes the transport), I've sometimes quipped they're doing it only to do cable manufacturers a favor, but whoever doesn't want to go that route but get a one-box dCS, had better get a Rossini or Bartok.

Greetings from Switzerland, David.
Or, the obvious, they can charge a whole helluva a lot more money than if they put it in one box.
 

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
People who scoff at multi-unit server/DACs tend to overlook the obvious: they have no problem adding an expensive audiophile server to their USB DAC, buy expensive USB cables etc. In our listening comparison of the different Lampizator models, for example, one of the things that struck me most is they not only thrive on, but need a dedicated server like the Lampizator Komputer. So there you have one type of multi-unit server/DAC with audiophile cable loom. Any modern dCS DAC is its own "server" (in the case of the Vivaldi, the Upsampler, which as mentioned above, is effectively an outhoused part of a server/DAC concept, is what's accessing files from external storage). Counting one fewer box(es). No use including the NAS one puts in the cellar or barn on that list, all server/DACs need access to file storage if, like me, one has nearly 20TB worth of digital files to play back.

This is particularly ironic since, adding a server and/or computer, they're adding to "computer audio" what I regard as the source of (almost) all evil they then pay top dollar to root out…

Greetings from Switzerland, David.

And why do you think that is? I am sure that if you measured the jitter of each of these servers and cables you would find a big part of the answer.
 

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
I have ignored nothing you have said. Please show where I am ignoring what you are saying. But you essentially disproved what you wanted to say by introducing a DAC, we both consider to be far from SOTA, into the discussion. You are proving my point, which you seem not to acknowldege but said back in post 910, that this waveform measurement means nothing...as in ZERO correlation with SQ.
The waveform reconstruction does have a relation to SQ. I'm repeating myself: if that measurement has not relation to SQ, then no measurement has. In short, I was proving my point.

Ignoring what I said: https://www.whatsbestforum.com/threads/state-of-the-art-digital.29583/post-691691

Positive versus negative correlation: think of it as a one-way street when it comes to measurements. I never ever claimed it goes both ways, you continue to argue as if I did. That a DAC measures well doesn't tell us much about SQ. That doesn't mean measurements don't prove anything, only that engineers haven't it all figured out yet. They'll eventually come up with more measurements that tell us at least something about a device's potential SQ. Just to remind you, many such measurements already exist, such as frequency response. Waveform reconstruction, being a measurement in the analogue domain, is another one.

As an audiophile, I regularly accuse my engineer friends of a lack of intellectual and hence scientific curiosity, in which they differ from my audiophile friends who are sometimes lacking in scientific and hence intellectual curiosity. How else could we be leading discussion in which consumers scoff at engineering driven companies when what happens, regardless of whether we like or buy their products, is that their inventions trickle down to products the same people have no qualms using. For example, anyone listening to DSD with a USB DAC (by any company building such a device) should consider writing the engineering team at dCS a Thank-You letter.

In contrast, jitter bugs, reclockers etc. do not address fundamental problems. Yes, one may prefer the sound of an aftermarket's product jitter spectrum, and refer to it being quantitatively lower as being the reason it sounds different, but one certainty in all this is that the jitter spectrum is now at least partially that of the jitter bug, reclocker, whatnot. These are devices that fall into the category of a solution looking for a problem. Where on the scale of positive versus negative correlation would you say such a device figures?

Greetings from Switzerland, David.
 
Last edited:
  • Like
Reactions: BMCG

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
Or, the obvious, they can charge a whole helluva a lot more money than if they put it in one box.
Indeed. Needless to say, they do offer alternatives, so do other companies, and it's not as if anyone has to buy into this philosophy. It's merely their attempt at addressing and solving problems fundamentally. The way I see it, what and how much of the inventions and expertise trickles down into affordable products is what's going to be important to the audiophile community.

Greetings from Switzerland, David.
 

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
And why do you think that is? I am sure that if you measured the jitter of each of these servers and cables you would find a big part of the answer.
Of course. That doesn't mean a USB DAC with dedicated server as a system is not what I alluded to before: a solution looking for a problem. I'd rather spend my money where problems are being addressed and solved fundamentally. Forgot to mention above, I know full well easier money is to be made with aftermarket products. But as an customer, I do not feel it's my responsibility to support this kind of philosophy. As an audiophile, I may only want great sound, as a pragmatist I'll still choose one great-sounding product over another taking such considerations into account.

By the way, speaking of money: does everyone realize those audiophile servers plus USB cables often cost as much as a Vivaldi Upsampler, Rossini DAC, easily as much as a Bartok DAC, all of which replace/do away with either?

Greetings from Switzerland, David.
 
Last edited:

the sound of Tao

Well-Known Member
Jul 18, 2014
3,620
4,838
940
The DACS being discussed and defended here may not be considered SOTA , as in using the latest technologies available. They are more evolutionary than a revolutionary. In the case of the Yggdrasil and others, they are a culmination of years based on an older proven design, proven to sound better than a DS type DAC. I consider it “SOTA” when a designer/engineer uses components designed for accuracy instead of using an off the shelf audio solution, instead using new software and methods taking advantage of the precision devices to deliver accurate sound.
I view the Porsche 911 in the same way.. An Old design constantly being tweaked and modified through the years to make the car perform like SOTA, Yet being outdated in many respects. Is that not delivering SOTA performance ?
On a personal level, I am enjoying my 16/44 redbook CD’s and highres formats thanks to an older design that has been carefully constructed to carry the original digital information intact from Beginning to end of the digital process.. whether that is considered SOTA or not, it is the process I prefer and delivers a sound I enjoy.
Much of the equipment discussed here is beyond my means and I have Not had The pleasure to experience. I have no doubt that the MSB select II DAC and others are phenomenal sounding to a level I may not be able to comprehend.
I guess some go about defining SOTA by the way it’s allowing them to enjoy digital music that doesn’t sound like 1’s and 0’s, that would be me.
I’m with you that it’s what we enjoy that is important and not what is currently Sota.

It is the kind of thing I’ve not necessarily aspired to and such a subjective call at any rate... but some things at some time do seem to just fit the bill of state of the art.

I’d really imagine Taiko’s Extreme is possibly a good example. Seemingly fits the criteria in areas of levels of performance and also providing innovation in technology. Digital is such a changing thing that the target there for Sota is so crazily quickly moving.

There are the things like a 911 that become classic and kind of move past the whole topic into something elevated beyond the immediate Sota.

But it’s not a title I spend much time thinking about but figure it is interesting because it seems to represent the moving culminating plateau in both technology and performance (even if these things are loosely enmeshed but not then necessarily correlated exactly).
 
Last edited:
  • Like
Reactions: Mikem53

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
The waveform reconstruction does have a relation to SQ. I'm repeating myself: if that measurement has not relation to SQ, then no measurement has. In short, I was proving my point.

Ignoring what I said: https://www.whatsbestforum.com/threads/state-of-the-art-digital.29583/post-691691

Positive versus negative correlation: think of it as a one-way street when it comes to measurements. I never ever claimed it goes both ways, you continue to argue as if I did. That a DAC measures well doesn't tell us much about SQ. That doesn't mean measurements don't prove anything, only that engineers haven't it all figured out yet. They'll eventually come up with more measurements that tell us at least something about a device's potential SQ. Just to remind you, many such measurements already exist, such as frequency response. Waveform reconstruction, being a measurement in the analogue domain, is another one.

As an audiophile, I regularly accuse my engineer friends of a lack of intellectual and hence scientific curiosity, in which they differ from my audiophile friends who are sometimes lacking in scientific and hence intellectual curiosity. How else could we be leading discussion in which consumers scoff at engineering driven companies when what happens, regardless of whether we like or buy their products, is that their inventions trickle down to products the same people have no qualms using. For example, anyone listening to DSD with a USB DAC (by any company building such a device) should consider writing the engineering team at dCS a Thank-You letter.

In contrast, jitter bugs, reclockers etc. do not address fundamental problems. Yes, one may prefer the sound of an aftermarket's product jitter spectrum, and refer to it being quantitatively lower as being the reason it sounds different, but one certainty in all this is that the jitter spectrum is now at least partially that of the jitter bug, reclocker, whatnot. These are devices that fall into the category of a solution looking for a problem. Where on the scale of positive versus negative correlation would you say such a device figures?

Greetings from Switzerland, David.
And I have made it clear that I don’t agree with the premise that the waveform correlates with sound quality and I also don’t think it means no measurements COULD correlate with SQ.
 

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
Of course. That doesn't mean a USB DAC with dedicated server as a system is not what I alluded to before: a solution looking for a problem. I'd rather spend my money where problems are being addressed and solved fundamentally. Forgot to mention above, I know full well easier money is to be made with aftermarket products. But as an customer, I do not feel it's my responsibility to support this kind of philosophy. As an audiophile, I may only want great sound, as a pragmatist I'll still choose one great-sounding product over another taking such considerations into account.

By the way, speaking of money: does everyone realize those audiophile servers plus USB cables often cost as much as a Vivaldi Upsampler, Rossini DAC, easily as much as a Bartok DAC, all of which replace/do away with either?

Greetings from Switzerland, David.
Maybe those other solutions sound better than the DCS solutions? Can those stream Qobuz or Tidal?
 

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
And I have made it clear that I don’t agree with the premise that the waveform correlates with sound quality and I also don’t think it means no measurements COULD correlate with SQ.
If that were true, then your premise is there are no measurements that correlate with sound quality (with the possible exception of frequency response perhaps, but maybe you think that's no indication of sound quality either), given there are few relevant measurements in the analogue domain of a DAC. Measurements in the digital domain are a different matter: correlation is yet harder to pin down. Of course, if that isn't what you meant to say, i.e. if you do think some measurements may be indicators of performance (in the positive sense) versus merely indicators of potential flaws, then by necessity, the ability of a converter to reconstruct an analogue waveform would be one of those. It's a digital-to-ANALOGUE converter after all.

Greetings from Switzerland, David.
 
Last edited:

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
Maybe those other solutions sound better than the DCS solutions? Can those stream Qobuz or Tidal?
They all can, no difference in that respect.

My point is simply this: how come the peanut gallery think it's hilarious when one says a dCS Vivaldi DAC without Upsampler won't sound its best (= 1 additional box and expense of ca. 20k), but find it perfectly acceptable that a USB DAC needs an audiophile server to sound its best (1 additional box and expense of 10-20k-plus depending on the choice of server)?

I'd not recommend either "without", that's all. Admittedly, if said USB DAC were, for example, a Lampizator Atlantic TRP, Golden Gate or Pacific, the need for a dedicated audiophile server is greater than for e.g. a Playback Designs Merlot because of superior USB implementation in some DACs than others, even so, I'd not want to plug in a Mac Mini or laptop for any USB DAC I've ever heard and call that a system. People who own an MSB Select II are going to tell you the same thing. Is this in any way more dubious as an assessment because of its price tag? Not to me it isn't.

There is no arguing taste. By all means, take your pick. But prospective buyers need to realize that if their audio dealer recommends a Vivaldi DAC without Upsampler, or a Lampizator Pacific without dedicated audiophile server, he's either an ignorant or a crook, as he should know they'll eventually want that extra device/unit.

That's why I thought it fair to point out that I wouldn't want to use a standalone Vivaldi DAC unit (at least not for RBCD playback). I'd say the exact same thing to anyone who wants to buy e.g. a Lampizator Pacific (regardless of the format one intends to play back), and likely most USB DACs. These are source component systems, be prepared to go the full monty.

Greetings from Switzerland, David.
 
Last edited:

the sound of Tao

Well-Known Member
Jul 18, 2014
3,620
4,838
940
They all can, no difference in that respect.

My point is simply this: how come the peanut gallery think it's hilarious when one says a dCS Vivaldi DAC without Upsampler won't sound its best with RBCD playback (= 1 additional box and expense of ca. 20k), but find it perfectly acceptable that a USB DAC needs an audiophile server to sound its best (1 additional box and expense of 10-20k-plus depending on the choice of server)?

I'd not recommend either "without", that's all. Admittedly, if said USB DAC were, for example, a Lampizator Atlantic TRP, Golden Gate or Pacific, the need for a dedicated audiophile server is greater than for e.g. a Playback Designs Merlot because of superior USB implementation in some DACs than others, even so, I'd not want to plug in a Mac Mini or laptop for any USB DAC I've ever heard and call that a system. People who own an MSB Select II are going to tell you the same thing. Is this in any way more dubious as an assessment because of its price tag? Not to me it isn't.

There is no arguing taste. By all means, take your pick. But prospective buyers need to realize that if their audio dealer recommends a Vivaldi DAC without Upsampler, or a Lampizator Pacific without dedicated audiophile server, he's either an ignorant or a crook, as he should know they'll eventually want that extra device/unit.

That's why I thought it fair to point out that I wouldn't want to use a standalone Vivaldi DAC unit (at least not for RBCD playback). I'd say the exact same thing to anyone who wants to buy e.g. a Lampizator Pacific (regardless of the format one intends to play back). These are source component systems, be prepared to go the full monty.

Greetings from Switzerland, David.
Peanut gallery comment aside... I’m not sure that in a digital front end the needs for a server and an upscaler are in the same boat. A server is a transport and would be widely considered essential infrastructure these days... for baseline fundamentals, whereas an upscaler would be more viewed as an add on tweak... perhaps more icing on the cake. One is essential and the other could be optional but just shouldn’t be necessary for at least baseline performance... unless DCS then package their totl dac with an upscaler as a two box unit and not market their premium dac as standalone and Sota.
 
Last edited:

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
I’m not sure that in a digital front end the needs for a server and an upscaler are in the same boat. A server is a transport and would be widely considered essential infrastructure these days... a baseline fundamental, whereas an upscaler would be more viewed as an add on tweak... perhaps more icing on the cake. One is essential and the other could be optional but just shouldn’t be necessary for baseline performance... unless DCS then package their totl dac with an upscaler as a two box unit and not just sell their premium dac as standalone.
I'm not saying one is the same as the other. Fact is, one effectively replaces the other, so from a financial perspective, one only needs to do the math. Most USB DACs I've heard so far thrive on or need a dedicated audiophile server. The three modern dCS DACs don't need one. If one wants a one-box solution because the Vivaldi is too expensive, takes up too much space, is too much of a hassle etc., get a Rossini or Bartok instead. If it's not too expensive and one still wants only one box, get Vivaldi One, which even includes the SACD transport.

As to several boxes versus one from an engineering perspective: several SOTA DAC systems come in multiple boxes, CH Precision, dCS, MSB etc. One could list arguments in favor and against multiple boxes. Who cares? No one's forced to buy multiple boxes.

The least one does need for file playback is a NAS for file storage, unless one is one of those audiophile who continually listen to the same handful albums or tracks (not me). That plus something to play back those files. What I'm questioning in this scenario is the need for a computer or server as it's technically speaking the source of those problems it purports to solve. Of course it can be made to not be in the way of great sound. But it wouldn't, if it weren't there in the first place.

Greetings from Switzerland, David.
 

the sound of Tao

Well-Known Member
Jul 18, 2014
3,620
4,838
940
I'm not saying one is the same as the other. Fact is, one effectively replaces the other, so from a financial perspective, one only needs to do the math. Most USB DACs I've heard so far thrive on or need a dedicated audiophile server. The three modern dCS DACs don't need one. If one wants a one-box solution because the Vivaldi is too expensive, takes up too much space, is too much of a hassle etc., get a Rossini or Bartok instead. If it's not too expensive and one still wants only one box, get Vivaldi One, which even includes the SACD transport.

As to several boxes versus one from an engineering perspective: several SOTA DAC systems come in multiple boxes, CH Precision, dCS, MSB etc. One could list arguments in favor and against multiple boxes. Who cares? No one's forced to buy multiple boxes.

The least one does need for file playback is a NAS for file storage, unless one is one of those audiophile who continually listen to the same handful albums or tracks (not me). That plus something to play back those files. What I'm questioning in this scenario is the need for a computer or server as it's technically speaking the source of those problems it purports to solve. Of course it can be made to not be in the way of great sound. But it wouldn't, if it weren't there in the first place.

Greetings from Switzerland, David.
I’d stand by what I said... it’s just different expectations about the base levels of performance that we perhaps should be getting in gear at this price level.

Also what we are talking about are just opinions and preferences on both approaches and outcomes but if I wanted Sota I’d not be pursuing a Bartok or a Rossini... and the Vivaldi has its own great characteristics but not all are chasing the same outcomes and I’d tend to look outside the DCS ecosystem given simply different preferences for different sonic qualities and music outcomes. That is the game in truth... finding the version of the best for us.
 

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
I’d stand by what I said... it’s just different expectations about the base levels of performance that we perhaps should be getting in gear at this price level.

Also what we are talking about are just opinions and preferences on both approaches and outcomes but if I wanted Sota I’d not be pursuing a Bartok or a Rossini... and the Vivaldi has its own great characteristics but not all are chasing the same outcomes and I’d tend to look outside the DCS ecosystem given simply different preferences for different sonic qualities and music outcomes. That is the game in truth... finding the version of the best for us.
No disagreement whatsoever. I also stand by what I said: if one were to get e.g. a dCS Vivaldi, be sure to get the Upsampler along with it to get the most out of it. If one were to get e.g. the Lampizator Pacific, be sure to get a dedicated audiophile server along with it to get the most out of it. If one were to get another USB DAC, be sure to at least try how much of a difference a dedicated audiophile server makes - before making a buying decision.

Greetings from Switzerland, David.
 
  • Like
Reactions: the sound of Tao

Al M.

VIP/Donor
Sep 10, 2013
8,681
4,470
963
Greater Boston
No disagreement whatsoever. I also stand by what I said: if one were to get e.g. a dCS Vivaldi, be sure to get the Upsampler along with it to get the most out of it. If one were to get e.g. the Lampizator Pacific, be sure to get a dedicated audiophile server along with it to get the most out of it. If one were to get another USB DAC, be sure to at least try how much of a difference a dedicated audiophile server makes - before making a buying decision.

Greetings from Switzerland, David.

Actually, for the networking/server capabilities you don't need the Upsampler. The dCS Network Bridge will suffice, and it is much cheaper. The Upsampler has extra functions (hey, upsampling!) that are independent of the server capabilities.
 

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
Actually, for the networking/server capabilities you don't need the Upsampler. The dCS Network Bridge will suffice, and it is much cheaper. The Upsampler has extra functions (hey, upsampling!) that are independent of the server capabilities.
Doesn't sound as good. I know someone who's gone that route initially, and there are people on this board who've compared and came to the same conclusion. Also, remember this whole discussion came up when ack mentioned RBCD playback with a standalone Vivaldi DAC without the Upsampler. It's a great unit regardless, but if one is going to use an Upsampler at all, then surely for RBCD playback (not saying it's not worth it for anything else, on the contrary, but picking just one upsampling application, that would be the one).

Greetings from Switzerland, David.
 
Last edited:

morricab

Well-Known Member
Apr 25, 2014
9,391
4,986
978
Switzerland
If that were true, then your premise is there are no measurements that correlate with sound quality (with the possible exception of frequency response perhaps, but maybe you think that's no indication of sound quality either), given there are few relevant measurements in the analogue domain of a DAC. Measurements in the digital domain are a different matter: correlation is yet harder to pin down. Of course, if that isn't what you meant to say, i.e. if you do think some measurements may be indicators of performance (in the positive sense) versus merely indicators of potential flaws, then by necessity, the ability of a converter to reconstruct an analogue waveform would be one of those. It's a digital-to-ANALOGUE converter after all.

Greetings from Switzerland, David.
No, that is not my premise at all. Please stop reading things into what I am saying that I am not saying. For example, I am saying that I think Jitter measurements COULD (if interpreted correctly) give insight into sound quality. So could distortion measurements (which for digital are a mix of digital artifacts and analog output stage distortions)...but again it comes down to how to interpret them. Just looking at raw plots and saying you can see how one will sound better than another is not the way.

For amplifiers (or more generally analog electronics), there has been more research into the correlation of preference to measurement. STarting way back with the BBC, D.E.L Shorter was trying to correlate the harmonic distortion with perceived sound quality. He came up with a relatively simplistic metric. That metric was a good start but incomplete. More recently Earl Geddes and Daniel Cheever both came up with metrics that take into account the nature of distotion and the nature of how we hear. Cheever, for example, found that our sensitivity is level dependent so his metric takes this and things like masking into account. You might be interested to know that Geddes actually found that the correlation with THD and IMD with sound quality was essentially zero but in fact was slightly negative! It just demonstrated that bulk numbers of distortion do not at all tell the story and in fact mislead. The nature of the distortion is far more important than the quantity of distortion...I suspect that this will be doubly true with digital. This is, to a large extent, why I think there are many people who think a properly implemented Philips TD1541 triple crown 16 bit ladder chip is still SOTA.

For me, a good analogy is with digital cameras. I was shopping once for one for my wife. Many of them had 12, 15 or even 20 million Mpixels. Then I found a Nikon, bottom of the range for SLR cameras, for about the same price as some of these others (which were more lifestyle cameras) that had "only" 6Mp. I asked the sales guy why this camera with far less Mp and almost no features was the same or more expensive. He simply replied "Becuase it takes really good pictures". So I bought it and he was absolutely correct. It took phenomenal pictures and that was all down to other factors than the number of Mp. Digging deeper, it was clear that the superior lens and the NATURE of the detector (it was a proper CCD and not a cheaper CMOS that is a lot noisier) made up for whatever limited pixels it might have.

To date, I have not seen anyone try to tackle correlation of measurements and sound quality of digital. Perhaps because digital processes and distortions are difficult to understand (like what effect does a filter that PRE-rings have on sound? What is pre-ringing in the real world? Obviously nothing in nature rings before the inital sound that make the ringing) and they are wholly unnatural and not found in our evolutionary past. It's even worse than the already unnatural harmonic distortion patterns made by most amps. Digital combines strange unnatural distortion types with the usual suspects in analog electronics. This makes unraveling the puzzle very complicated.

This is why everyone touting "perfect" measurements lead to the best sound should be even more careful about such proclamations than those who in previous decades claimed this about analog electronics, because the correlation is currently not at all established. It is also why your claim that if the waveform doesn't mean anything then none of the measurements do. You couldn't possibly know this first of all and there are a lot of other things that could be causing it rather than such a macro measurement like waverform profile. You should probably stop talking to engineers. They know how to design the stuff but very few of them know what they are trying to design to...so they design to achieve lowest distortion, jitter etc. without thinkng about how they get there. Scientific research on the correlation of sound to measurement should be the guide but if followed leads a lot of engineers into an uncomfortable and vague space where they can't just engineer to the lowest of everything like they can with a RADAR. Engineering to a nebulous "human preference" is in many ways a lot tougher.
 

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
No, that is not my premise at all. Please stop reading things into what I am saying that I am not saying. For example, I am saying that I think Jitter measurements COULD (if interpreted correctly) give insight into sound quality. So could distortion measurements (which for digital are a mix of digital artifacts and analog output stage distortions)...but again it comes down to how to interpret them. Just looking at raw plots and saying you can see how one will sound better than another is not the way.

For amplifiers (or more generally analog electronics), there has been more research into the correlation of preference to measurement. STarting way back with the BBC, D.E.L Shorter was trying to correlate the harmonic distortion with perceived sound quality. He came up with a relatively simplistic metric. That metric was a good start but incomplete. More recently Earl Geddes and Daniel Cheever both came up with metrics that take into account the nature of distotion and the nature of how we hear. Cheever, for example, found that our sensitivity is level dependent so his metric takes this and things like masking into account. You might be interested to know that Geddes actually found that the correlation with THD and IMD with sound quality was essentially zero but in fact was slightly negative! It just demonstrated that bulk numbers of distortion do not at all tell the story and in fact mislead. The nature of the distortion is far more important than the quantity of distortion...I suspect that this will be doubly true with digital. This is, to a large extent, why I think there are many people who think a properly implemented Philips TD1541 triple crown 16 bit ladder chip is still SOTA.

For me, a good analogy is with digital cameras. I was shopping once for one for my wife. Many of them had 12, 15 or even 20 million Mpixels. Then I found a Nikon, bottom of the range for SLR cameras, for about the same price as some of these others (which were more lifestyle cameras) that had "only" 6Mp. I asked the sales guy why this camera with far less Mp and almost no features was the same or more expensive. He simply replied "Becuase it takes really good pictures". So I bought it and he was absolutely correct. It took phenomenal pictures and that was all down to other factors than the number of Mp. Digging deeper, it was clear that the superior lens and the NATURE of the detector (it was a proper CCD and not a cheaper CMOS that is a lot noisier) made up for whatever limited pixels it might have.

To date, I have not seen anyone try to tackle correlation of measurements and sound quality of digital. Perhaps because digital processes and distortions are difficult to understand (like what effect does a filter that PRE-rings have on sound? What is pre-ringing in the real world? Obviously nothing in nature rings before the inital sound that make the ringing) and they are wholly unnatural and not found in our evolutionary past. It's even worse than the already unnatural harmonic distortion patterns made by most amps. Digital combines strange unnatural distortion types with the usual suspects in analog electronics. This makes unraveling the puzzle very complicated.

This is why everyone touting "perfect" measurements lead to the best sound should be even more careful about such proclamations than those who in previous decades claimed this about analog electronics, because the correlation is currently not at all established. It is also why your claim that if the waveform doesn't mean anything then none of the measurements do. You couldn't possibly know this first of all and there are a lot of other things that could be causing it rather than such a macro measurement like waverform profile. You should probably stop talking to engineers. They know how to design the stuff but very few of them know what they are trying to design to...so they design to achieve lowest distortion, jitter etc. without thinkng about how they get there. Scientific research on the correlation of sound to measurement should be the guide but if followed leads a lot of engineers into an uncomfortable and vague space where they can't just engineer to the lowest of everything like they can with a RADAR. Engineering to a nebulous "human preference" is in many ways a lot tougher.
Adding to all this, I'm seeing problems as measurements are invariably being taken in isolation, plus are often unspecific, e.g. while I have no doubt measuring analogue waveform reconstruction is relevant (after all, it's what a DAC does or is supposed to do, convert into analogue), it's merely one step towards proof that a converter that outputs perfect sine waves will accurately reconstruct complex waveforms (e.g. music).

My father used to work in a photo shop for several years, by the way, so I know what he'd tell you in reply to your story on digital cameras - that the camera that took better pics and cost more in all likelihood had a better lens. Basically the same type of argumentation that the Yggy users above pointed to, that the 20-bit or 24-bit (whatnot) resolution shouldn't be considered the bottleneck when e.g. the total system noise floor precludes the reproduction of 24-bit resolution.

The part I happen to find problematic in discussions like this (e.g. you mention THD and IMD) is that someone invariably hastens to add that human beings can't hear the difference anyway, pointing to studies whose results are ultimately based on negative evidence. Logically speaking, the absence of evidence isn't proof (I'm not religious, but you get the gist). The resulting "can't be" answer we audiophiles invariably get from engineering geeks isn't just lame, it's unscientific, and only points to a lack of intellectual curiosity.

As to engineering a product to human preference, I'm reminded of what a dear friend of mine who is a Federal Chef who teaches further education to junior chefs says about cooking: one is better off and bound to live a happier life cooking something one likes, crossing one's fingers that others will like it, too.

I have no doubt that is what the designers of the DACs we've discussed here lately do or are trying to do, with the possible exception of a Daniel Weiss, who scoffed at me once when I mentioned listening to one's own product - he was adamant about the objective nature of measurements, in short, he's convinced his products must sound good because they measure well, I got the impression he never listens to his products. It's not surprising to me that his best products are for professional studio applications, e.g. D/D converters, conversion software etc., all safely in the digital domain. His consumer products, I'd prefer to leave it at it when I don't have anything positive to say, as the bashing of products is a type of attitude I already find difficult to tolerate in discussion forums such as this.

More importantly, the existence of such rare birds in the audio business isn't proof engineers are all nerds or deaf or both, and even if they were, aren't employed where they contribute to the excellence of a product, being led by someone who has a philosophy and does listen. Without engineers, we'd not be here discussing audiophile products. The trap audiophiles fall into is really twofold: not everyone who listens builds great products (same as not everyone with taste buds is a great cook). We may not agree at which end of the spectrum more snake oil is being sold, but personally, I have just as much or little confidence in a designer who makes it a point to inform his clientele he listens as in one who says he measures - I let my ears decide what I like or don't like.

Greetings from Switzerland, David.
 
Last edited:

acousticsguru

Well-Known Member
Feb 17, 2014
505
324
373
No, that is not my premise at all. Please stop reading things into what I am saying that I am not saying. For example, I am saying that I think Jitter measurements COULD (if interpreted correctly) give insight into sound quality. So could distortion measurements (which for digital are a mix of digital artifacts and analog output stage distortions)...but again it comes down to how to interpret them. Just looking at raw plots and saying you can see how one will sound better than another is not the way.

For amplifiers (or more generally analog electronics), there has been more research into the correlation of preference to measurement. STarting way back with the BBC, D.E.L Shorter was trying to correlate the harmonic distortion with perceived sound quality. He came up with a relatively simplistic metric. That metric was a good start but incomplete. More recently Earl Geddes and Daniel Cheever both came up with metrics that take into account the nature of distotion and the nature of how we hear. Cheever, for example, found that our sensitivity is level dependent so his metric takes this and things like masking into account. You might be interested to know that Geddes actually found that the correlation with THD and IMD with sound quality was essentially zero but in fact was slightly negative! It just demonstrated that bulk numbers of distortion do not at all tell the story and in fact mislead. The nature of the distortion is far more important than the quantity of distortion...I suspect that this will be doubly true with digital. This is, to a large extent, why I think there are many people who think a properly implemented Philips TD1541 triple crown 16 bit ladder chip is still SOTA.

For me, a good analogy is with digital cameras. I was shopping once for one for my wife. Many of them had 12, 15 or even 20 million Mpixels. Then I found a Nikon, bottom of the range for SLR cameras, for about the same price as some of these others (which were more lifestyle cameras) that had "only" 6Mp. I asked the sales guy why this camera with far less Mp and almost no features was the same or more expensive. He simply replied "Becuase it takes really good pictures". So I bought it and he was absolutely correct. It took phenomenal pictures and that was all down to other factors than the number of Mp. Digging deeper, it was clear that the superior lens and the NATURE of the detector (it was a proper CCD and not a cheaper CMOS that is a lot noisier) made up for whatever limited pixels it might have.

To date, I have not seen anyone try to tackle correlation of measurements and sound quality of digital. Perhaps because digital processes and distortions are difficult to understand (like what effect does a filter that PRE-rings have on sound? What is pre-ringing in the real world? Obviously nothing in nature rings before the inital sound that make the ringing) and they are wholly unnatural and not found in our evolutionary past. It's even worse than the already unnatural harmonic distortion patterns made by most amps. Digital combines strange unnatural distortion types with the usual suspects in analog electronics. This makes unraveling the puzzle very complicated.

This is why everyone touting "perfect" measurements lead to the best sound should be even more careful about such proclamations than those who in previous decades claimed this about analog electronics, because the correlation is currently not at all established. It is also why your claim that if the waveform doesn't mean anything then none of the measurements do. You couldn't possibly know this first of all and there are a lot of other things that could be causing it rather than such a macro measurement like waverform profile. You should probably stop talking to engineers. They know how to design the stuff but very few of them know what they are trying to design to...so they design to achieve lowest distortion, jitter etc. without thinkng about how they get there. Scientific research on the correlation of sound to measurement should be the guide but if followed leads a lot of engineers into an uncomfortable and vague space where they can't just engineer to the lowest of everything like they can with a RADAR. Engineering to a nebulous "human preference" is in many ways a lot tougher.
What I really have on my mind in this "measurements versus listening" discussion is a) that neither of us appears to like DACs whose designers don't listen to, and b) that now that "thanks to" the pandemic, I have time to listen every night into the wee hours (4-5 a.m. almost every night this week alone), I'm loving it using a DAC brand you keep knocking, all the while you're touting DACs (by one brand in particular) that make me feel as if I were being pierced by ice rain. You'll notice I spend NO time on forums bashing those products, make of it, what you will…

Greetings from Switzerland, David.
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
My father used to work in a photo shop for several years, by the way, so I know what he'd tell you in reply to your story on digital cameras - that the camera that took better pics and cost more in all likelihood had a better lens. Basically the same type of argumentation that the Yggy users above pointed to, that the 20-bit or 24-bit (whatnot) resolution shouldn't be considered the bottleneck when e.g. the total system noise floor precludes the reproduction of 24-bit resolution.

IMHO we can not use analogies with cameras - they are a recording device, not a playing device. But you are correct - the quality of the lens is a main aspect in their performance. But then we have different opinions - what is the best "lens" - one that delivers the perfect optical transmission or one lens that manipulates the transmission to get the best picture according to people who see the photos? Also, contrary to what people can imagine, getting an high-resolution picture is an extremely complex process - far from simply collecting the photo-electrons in a well! Nowadays we can not compare SOTA lenses independently of the camera.

BTW, the Nikon-versus-Canon-lens-debate is sometimes even more spirited than our DAC debates in WBF!
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing