[Joe Bloggs]

Strictly speaking that is incorrect...

The original version of USM was done optically, by superimposing a negative image that was slightly blurred over the original, and optically summing them, so it literally applied a halo only at edges of high contrast... (At low contrast edges, while the falloff profile of the edge may be affected, it won't be boosted or cut enough to appear as an actual halo. Apparently, while changing the dropoff profile may be slightly visible, only an actual halo produces a distinct illusion of artificially added sharpness.) The blurring applied to the "mask image" is where the term "unsharp masking" comes from.

Keith...

The "slightly blurred negative image" = lowpassed image with inverted phase

Optically summing them = cutting all frequencies except those cut by the lowpass

Further lighting up the summation = boosting the frequency bands cut by the lowpass.

And no, "an actual halo" is not the preferred result. If so why don't you add some actual halos to your perfectly sharp LCD monitor using the sharpness control that may or may not be there anymore?

Modern digital versions produce a similar effect by signal processing, but virtually all of them apply the effect ONLY TO EDGES, after using some method to avoid applying it to "non-edges". Therefore, if you want a more correct analogy, they act more like a dynamic processing equalizer, which is applied unequally and whose effect varies with time and the specific characteristics of the signal it is applied to.

On a typical DAC...
- If you pass a continuous sine wave of fixed frequency through them there will be no measurable effect.
- If you pass a variety of frequencies, like a frequency sweep, the filters will have a minor effect on the overall frequency response (outside of its intended range).
- But, if you apply a TRANSIENT signal, then the filter will introduce ringing, which will vary greatly depending on the specific filter parameters used.
(For example, if you play a five second burst of a sine wave, the central portion will be virtually unaffected, but ringing will be introduced that extends PAST THE ENDS of the signal.)

The effect of the interpolation filter is always there.
1. If there's no interpolation filter you get no sound.
2. If a non-oversampled zeroth order hold is used, high frequencies pure tones will sound very funky indeed (like a square wave of the frequency f modulated by a Nyquist-2f beat frequency, AFAICT). But a transient signal won't ring!
3. If a proper filter is introduced, it is close to as you say, with respect to their analog originals. With the exception of the last point: within the confines of sampling theory, there's no such thing as a transient signal that ought to appear as input to a DAC. A sine wave start starts and stops on a dime, or a Kronecker delta, are artificial constructs that can never appear in a real-life recording, because they are cropped versions of theoretical signals that contain frequencies extending beyond Nyquist out to infinity. The "introduced ringing" is simply what happens when these theoretical signals are brought back to reality, rendered using the finite set of frequencies available under the sampling format.

Speaking of which, there IS a theoretically perfect reconstruction filter against which all real life reconstruction filters are to be measured: the infinite length sinc filter. It will produce ringing tails before and after your sine burst and K-delta--infinitely long tails! But said ringing would only consist of frequencies a fraction of a Hz under 22.05kHz (for 44.1kHz sample rate) and hence be completely inaudible (to you and me anyway)!

As noted in a previous post--the *shorter* the ringing response to a K-delta, the more artifacted the filter's response to real life input--including music and high frequency sine tones!

Therefore, if you look at the effect in relation to time, it will have no effect on steady state sine waves.
And will have an effect that is greatest at and near the edges of CHANGES in the signal.

As noted before, the ONE WAY in which I could detect imperfect filtering in pedestrian DACs was by listening for aliasing in high frequency sine tones--NOT by listening to K-delta clicks, square waves, theoretical sine pulses, or anything like that!

Incidentally, in modern high quality cameras, there is rarely a significant "drop in high frequency response". Typically, in most high quality modern cameras I've used, the effect you're compensating for is "pixel uncertainty". (If a sharp boundary between black and white falls in the center of a camera pixel, that pixel will be recorded as grey. Applying a USM to that image does NOT restore the original resolution, which was never recorded. There's no way, after the fact, to determine whether that grey pixel "should have been half black and half white" rather than being a grey pixel. Rather, the "sharpening effect" creates an optical illusion of "enhanced sharpness" which "perceptually compensates" for the lack of real resolution. The picture is "made to appear sharper" - whether it was originally that way or not.)

There's an antialiasing filter in front of the sensor to prevent that scenario and to replace that with, exactly, "a drop in high frequency response". If not, the sensor is high resolution enough that lens aberrations take care of the antialiasing. If not, serious artifacts occur when shooting finely patterned fabrics.

--Joe Bloggs
--10+ year dpreview member
--designed sharpening-based filter for HDR toning that sharpened *everywhere in the image except where halos would occur*: https://www.dpreview.com/forums/thread/2902464#forum-post-36868711

 

[Joe Bloggs]

That is the core problem; you always insist that any difference in > 20 kHz, lower than so and so % THD, etc, performance has no effect on audible differences. If you can, try to have both micros side by side - an afternoon is more than it takes, regardless how you will compare them. It is the only case of a commercialy available device known to me where significantly better electronic parts - when taken together - clearly result in an audible improvement. The THD, etc numbers are crazy in any case - and are particularly off the charts with Black Label. It should not matter audibly - yet, it does. Because the SQ is largely determined by analogue stages - input in ADC, output in DAC.

It might answer the question " How transparent is transparent? " ...

Transparent is transparent, unless, as Keith prefers it, they pass a DBT of a Sabre DAC vs whatever, or SOMETHING is done past the current sighted "I think I hear an elevation in treble when I think there's no actual elevation in treble even though I haven't measured the particular DACs to see if there's an actual elevation in treble and no none of this is volume matched or blinded or any more than your usual subjective hoolabaloo except you should take me seriously because I'm a serious guy!" situation.

 

[Jaywalk3r]

That is the core problem; you always insist that any difference in > 20 kHz, lower than so and so % THD, etc, performance has no effect on audible differences. If you can, try to have both micros side by side - an afternoon is more than it takes, regardless how you will compare them. It is the only case of a commercialy available device known to me where significantly better electronic parts - when taken together - clearly result in an audible improvement. The THD, etc numbers are crazy in any case - and are particularly off the charts with Black Label. It should not matter audibly - yet, it does. Because the SQ is largely determined by analogue stages - input in ADC, output in DAC.

It might answer the question " How transparent is transparent? " ...

I've highlighted the core problem above. It very much matters how they are compared. If the comparison is done in such a way as to not eliminate biases, it doesn't matter what you think you hear. It sure is funny how frequently "night & day" differences disappear when cognitive bias is eliminated from the comparison process.

 

[bigshot]

" How transparent is transparent? "

I know the answer to that question very well. Transparent is the line where no human can hear differences any more. I've been reminded of the all-too-human frailty of my ears on many occasions. I've spent a great deal of time researching and testing to have a basic idea of where the thresholds of audibility lie. I know some people (present company included!) claim to have supersonic ears. I give those folks a chance to back up their claims of "night and day differences" and "clearly audible" and "not at all subtle". If it really is that obvious, it should be easy to demonstrate that difference to the rest of us. But when my basic questions get answered with obfuscating paragraphs packed with confusing association with causation, bandwagon effect, begging the question and appeal to ignorance arguments, I start to tune out and wait for my simple questions to be answered simply.

All I am asking for is an audible difference. That shouldn't be difficult to prove. I have dozens of amps and DACs and players here in all different price ranges and I've compared them all INCLUDING ONE WITH A HIGH END SABRE CHIP, and I have yet to find one that sounds different. So I ask if someone can point me to one that sounds different and I get "oh I don't own that DAC any more..." and "the difference won't show up in any measurements" and "the only way to know is to conduct elaborate controlled tests on 50 people and run the percentages". That's when my BS meter starts to rise.

I'm nice and polite and I give people the opportunity to share some new info with me... until they start making me jump through hoops and try to slip a noose of logical fallacies around my neck. I'm not stupid or gullible. Those tricks might work on audiophools, but not me. If a difference is clearly audible, it should show up in measurements and it should be capturable and it should be clearly demonstrable in gear the person who is making the claim actually owns or has available for testing.

I shouldn't have to put a grand on MY credit card to prove YOUR anecdotal and so far unverified claim. At some point I get tired of the circular arguments and just call for the "put up or shut up". If I still keep getting the hoop being held up and the whip cracking and the encouragement to "JUMP! JUMP!", I dismiss them with a wave of the hand and move on and I don't read much of their posts any more. I've already gotten to that point with a couple of people in Sound Science. You can take a guess if you are one of those folks or not.

It's put up or shut up time! Would someone like to suggest simple ways to objectively verify a clearly audible difference between two DACs? Yes? No?

 

[Joe Bloggs]

I know the answer to that question very well. Transparent is the line where no human can hear differences any more. I've been reminded of the all-too-human frailty of my ears on many occasions. I'm very familiar with not being able to tell a difference. I know some people (present company included!) claim to have supersonic ears. I give those folks a chance to back up their claims of "night and day differences" and "clearly audible" and "not at all subtle". If it really is that obvious, it should be easy to demonstrate that difference to the rest of us. But when my basic questions get answered with obfuscating paragraphs packed with confusing association with causation, begging the question and appeal to ignorance arguments, I start to tune out and wait for my simple questions to be answered simply.

All I am asking for is an audible difference. That shouldn't be difficult to prove. I have dozens of amps and DACs and players here in all different price ranges and I've compared them all, and I have yet to find one that sounds different. So I ask if someone can point me to one that sounds different and I get "oh I don't own that DAC any more..." and "the difference won't show up in any measurements" and "the only way to know is to conduct elaborate controlled tests on 50 people and run the percentages". That's when my BS meter starts to rise.

I'm nice and polite and I give people the opportunity to share some new info with me... until they start making me jump through hoops and try to slip a noose of logical fallacies around my neck. I'm not stupid or gullible. Those tricks might work on audiophools, but not me. If a difference is clearly audible, it should show up in measurements and it should be capturable and it should be clearly demonstrable in gear the person who is making the claim actually owns or has available for testing.

I shouldn't have to put a grand on MY credit card to prove YOUR anecdotal and so far unverified claim. At some point I get tired of the circular arguments and just call for the "put up or shut up". If I still keep getting the hoop being held up and the whip cracking and the encouragement to "JUMP! JUMP!", I dismiss them with a wave of the hand and move on and I don't read much of their posts any more. I've already gotten to that point with a couple of people in Sound Science. You can take a guess if you are one of those folks or not.

It's put up or shut up time! Would someone like to suggest simple ways to objectively verify a clearly audible difference between two DACs? Yes? No?

This guy on facebook claims to be able to DBT between two DACs even when he messes with the volume. Sounds promising even if obnoxious at first!
https://www.facebook.com/groups/headfi.org/permalink/10156926474584736/?comment_id=10156927508814736&reply_comment_id=10156927944254736&comment_tracking={"tn":"R"}

https://www.facebook.com/vlad.petric?fref=gc&dti=4623534735

 

[bigshot]

I'm very familiar with crackpots who jump the volume to incredible levels to detect noise floors. We used to have one around here that the minimum acceptable dynamic range for recorded music is 120dB. He didn't seem to realize that his human ears would incur hearing damage trying to hear all 120dB of that dynamic range. Performance in extreme situations isn't what I'm asking for. I'm asking for two DACs that sound different playing a typical music CD at a normal listening level. That is how I use my audio equipment.

Again, I know differences exist. I want real world AUDIBLE differences.

 

[Joe Bloggs]

I'm very familiar with crackpots who jump the volume to incredible levels to detect noise floors. That isn't what I'm asking for. I'm asking for two DACs that sound different playing a typical music CD at a normal listening level. That is how I use my audio equipment.

Again, I know differences exist. I want AUDIBLE differences.

It's not that, he's just saying that he doesn't even need to volume match the two DACs to detect a difference anymore. The obvious question would be how one blinds two DACs AND mismatch their volume, he claims he has his ways.

 

[bigshot]

So did our old friend from the Geek Squad. Some people invest their ego in trying to prove that they are superior beings. I'm quite satisfied with my perfectly normal human ears. I just want sound that is as good as my ears can hear. I don't need to split atoms.

I just noticed the latest two tags that were added to this thread... haha!

 

[500178]

o my, this topic is the embodiment of audiophilia.

 

[TheSonicTruth]

I'm very familiar with crackpots who jump the volume to incredible levels to detect noise floors. We used to have one around here that the minimum acceptable dynamic range for recorded music is 120dB. He didn't seem to realize that his human ears would incur hearing damage trying to hear all 120dB of that dynamic range. Performance in extreme situations isn't what I'm asking for. I'm asking for two DACs that sound different playing a typical music CD at a normal listening level. That is how I use my audio equipment.

Again, I know differences exist. I want real world AUDIBLE differences.

"We used to have one around here
that the minimum acceptable dynamic
range for recorded music is 120dB."

And who might that have been?

 

[KeithEmo]

I think we've been discussing "the proper way to do it"... for quite some time.
And, for quite some time, I've been bombarded with complaints that "I haven't provided test results to support my hypothesis".
You're quite correct; I haven't.... which leaves it unproven at this point.
That's why I'm calling it a hypothesis rather than a FACT.

If we're just comparing two products, then the proper way would be to conduct a double-blind test of those two products.
WIth a proper set of test equipment; and properly chosen test samples; and a properly sized and selected sample.
If anybody wants to do so I would be glad to participate (by helping to determine the proper way to ensure that we get valid results).
However, no, I am NOT personally interested enough to do it myself.
(And, somewhat sadly, Emotiva isn't likely to sponsor an expensive study that isn't likely to justify its cost by selling more product... sorry.)

If you REALLY want to conduct a test on those two specific products PERSONALLY, I believe you'll find both the Emotiva DC-1 and the Wyred4Sound DAC2 readily available used.
(Both have pretty good resale value, so you could probably buy both, test them, and sell one or both for what you paid for it.)
The Emotiva DC-1 is surrently out of production.... but its replacement (the DC-2) will be available shortly.
All of Emotiva's products come with an unconditional 30 day return policy - which offers anyone who wants to the opportunity to do any sort of comparison they like.
(And, if someone wants to sponsor a real proper scientific study, with proper controls, associated equipment, and samples, I can quite possibly get you a loaner.)

Also, PLEASE, stop incorrectly putting words in my mouth....
I did NOT say "that the differences won't show up in any measurement".
I said that I suspect they will not show up in one particular measurement (frequency response); and I also conceded that I hadn't confirmed that.
I also noted that there are in fact CLEARLY measurable differences - and I described in which measurement they can be easily measured.
And I suggested my hypothesis - which is that one particular difference will correlate to audible differences.
Of course, even if there are confirmed audible differences, I many be incorrect about the cause.
(But YOU choose to insist that "differences in that particular measurement cannot possibly be audible and so cannot POSSIBLY be the cause"... )
(I'm willing to wait for actual test results - when and if we have any...)

Let me also note, yet again, that for the most part I agree with your claims...
There is FAR too much snake oil out there...
(And, yes, some of the differences that really do exist are actually rather trivial and probably not "important".)

------------------------------------------------
(Consider everything from here on to be "background information" - so you don't have to read it.)

If we are talking about how to do "a properly run and scientifically valid test"...
One difficulty is going to be to determine what constitutes a sufficient number of test subjects.
If it turns out that only one person in twenty can hear the difference, then we'll need a test sample of a few hundred to avoid missing those few, and producing a "false negative".
Assuming we're doing a SCIENTIFIC study, then saying "most people" really isn't good enough.
(I DON'T NOT believe this would be the case... since , of the many people I've asked, MOST of them calim to hear a similar difference to what I do... but I cannot rule it out.)

Another difficulty is that we must ENSURE that both our test equipment, and our test samples, do not obscure what we're testing for.
So, for example, our samples, and our speakers, must be able to reproduce the differences in waveforms caused by the differences in the various filters.
If we can see a difference in the output waveforms on an oscilloscope then we must make sure it is being delivered to the ears of our test subjects.
This is a common failure of many tests; they fail to ensure that their test setup can actually accurately deliver the differences they're claiming to be trying to measure.
If you're attempting to prove, or rule out, that people can hear 45 kHz, then your first step is to make sure it is present in your test samples, and that the speakers you use can reproduce it.
(And, when you heard no differences between those DACs, DID you verify that the equipment you used was accurately reproducing the output waveforms?)

And, if you want to prove a GENERALIZATION (about "all DACs" or "Sabre vs other DACs") rather than about two particular products, then you'll need to expand your test considerably.
Testing two specific products is very limited; especially when both have probably been through many minor design revisions... some of which may affect how they sound or measure.
In fact, you should probably use several smaples from different production runs, just to rule out "unit to unit production variations".

However, in this case, I am not aware of ANYONE who has conducted a proper "credible" study of whether the differences in DAC filters are audible.
ESS, who makes the Sabre DAC, claims to have "run focus groups" and to have "used the filter which most people preferred" (I take that as a claim to have found a difference).
I should also note that, when their product was introduced, they were introducing a "premium chip" (more expensive) - so they HAD to claim that it was DIFFERENT to justify the price.
Many other DAC manufacturers offer a choice of a variety of filters (we'll never know if they did any audibility tests or are simply offering a feature their customers have asked for).
Yet many other DAC manufacturers offer only a single filter (which, since they all cost about the same, suggests they believe the one they chose is the most accurate).

I should also point out that many so-called "reliable sources" seem to not stick to their story.
Sony released the CD format - based on the claim that "CDs sounded audibly identical to the original source to most people".
They then followed up a few years later by releasing the SACD format - which "corrected" many of the "audible limitations" of their "audibly perfect CD format".
So, WHICH Sony claim would you prefer to consider credible?

Many people claim to hear obvious differences between DACs; and many claim not to; so all we have there is a whole bunch of anecdotal information.

I am NOT suggesting that you do or don't buy an Emotiva DAC or a Wyred4Sound DAC.
Although, of course, if you believe that they sound the same, ours was a better deal because it cost less. (But, then, you can buy a decent sound card for $25).
You asked me to offer a specific example of where I heard a significant and obvious difference between DACs; I answered your question.
I chose Sabre DACs because I find the audible differences between them and most others rather obvious.
(And, no, I tend not to hang onto equipment I don't especially like - just to PROVE to other people why I don't like it.)

I will, however, note that I have had a lot of conversations with a lot of people about "how DACs sound" (since part of my job is providing support for customers who purchase DACs).
In find that roughly somewhat more than half of the people I talk to, if asked, agree that Sabre DACs "seem to have a distinctive sound".
Furthermore, OF THOSE WHO CLAIM TO HEAR A DIFFERENCE, almost all of them describe the difference they think they hear quite similarly, whether they like or dislike it.
(Those who like it say that they find Sabre DACs to be "more detailed"; those who dislike it consistently describe Sabre DACs as sounding "more grainy" or "etched".)
Feel free to consider this to be "a sociaological study" if you like... but the majority of people who believe they hear a difference seem to believe they hear a SIMILAR difference.
(And, oddly, it is much the same difference that I think I hear.)

Personally, I think it would be interesting to run a proper study,
I doubt that my company will ever do so... because it really is unlikely to help us sell more products.

As I noted in another post... there is very little "pure science" left in the world - and virtually none when it comes to audio equipment.

A company who sells $500 interconnects has little incentive to do a proper study (because it will probably show that their product is eitehr snake oil or only marginally better).
Yet the company who sells $5 interconnects ALSO has little incentive to sponsor the test.
(They'd have to sell an awful lot more $5 cables to justify the cost of a study; and most of their customers simly don't care anyway.)

And, for those who keep on trotting out studies, like the one conducted by Sony, that purport to "prove" that 16/44k audio is "audibly indistinguishable from the orignal...
I would first point out that Sony did NOT conduct an "unbiased scientific study" there.
They figured out the format they preferred, based on how much time they could get on a disc, and the current technology, then conducted a study to CONFIRM that it was "good enough".
(They had a major bias AGAINST determining that they COULDN'T produce a disc at the time that would fit all their requirements.)

I know the answer to that question very well. Transparent is the line where no human can hear differences any more. I've been reminded of the all-too-human frailty of my ears on many occasions. I've spent a great deal of time researching and testing to have a basic idea of where the thresholds of audibility lie. I know some people (present company included!) claim to have supersonic ears. I give those folks a chance to back up their claims of "night and day differences" and "clearly audible" and "not at all subtle". If it really is that obvious, it should be easy to demonstrate that difference to the rest of us. But when my basic questions get answered with obfuscating paragraphs packed with confusing association with causation, bandwagon effect, begging the question and appeal to ignorance arguments, I start to tune out and wait for my simple questions to be answered simply.

All I am asking for is an audible difference. That shouldn't be difficult to prove. I have dozens of amps and DACs and players here in all different price ranges and I've compared them all INCLUDING ONE WITH A HIGH END SABRE CHIP, and I have yet to find one that sounds different. So I ask if someone can point me to one that sounds different and I get "oh I don't own that DAC any more..." and "the difference won't show up in any measurements" and "the only way to know is to conduct elaborate controlled tests on 50 people and run the percentages". That's when my BS meter starts to rise.

I'm nice and polite and I give people the opportunity to share some new info with me... until they start making me jump through hoops and try to slip a noose of logical fallacies around my neck. I'm not stupid or gullible. Those tricks might work on audiophools, but not me. If a difference is clearly audible, it should show up in measurements and it should be capturable and it should be clearly demonstrable in gear the person who is making the claim actually owns or has available for testing.

I shouldn't have to put a grand on MY credit card to prove YOUR anecdotal and so far unverified claim. At some point I get tired of the circular arguments and just call for the "put up or shut up". If I still keep getting the hoop being held up and the whip cracking and the encouragement to "JUMP! JUMP!", I dismiss them with a wave of the hand and move on and I don't read much of their posts any more. I've already gotten to that point with a couple of people in Sound Science. You can take a guess if you are one of those folks or not.

It's put up or shut up time! Would someone like to suggest simple ways to objectively verify a clearly audible difference between two DACs? Yes? No?

A fact often exploited by snake-oil salesman. Audiophiles are, collectively, a pretty gullible bunch.

When it comes to high-end audio, snake-oil is the null hypothesis.

And no, if I can plug some high sensitivity, low impedance IEMs into a DAP, pause the music, set the gain to high and crank up the volume, and hear no noise, I don't care that the THD is 0.1% instead of 0.0003%. Sure, the difference is significant, but the effect isn't of a meaningful magnitude. Research suggests it isn't audible. Why would I pay extra? If a company wants me to pay for improved performance, they must first convince me that the increased performance adds value. If their claims conflict with prevalent scientific research, you'd better believe they better have some sound scientific research supporting those claims.

I'm not looking for the "why" of it yet. I'm skipping over that part. We'll get to the "why" of it later. I'm also not going to consider the reputation of the company. That is irrelevant to what we are doing here.

I'm looking for verification of the difference you heard. If there is a boost of at least 1.5dB between 5 and 7kHz which was your estimation of the difference, that would be reflected in the response. (If this isn't an accurate description of the difference, perhaps you can elaborate on what you heard better so we can understand better.) Distortion can be tested for as well. If the difference isn't reflected in the specs, either 1) the specs on one or both of the DACs is fudged/incomplete or 2) the particular DAC you compared was performing out of spec. I suppose there could be some magical thing that can be heard but not measured, but that would be something to consider much further down the road after we've eliminated all the other possibilities. I have been swamped at the studio this week. I'll google specs and bring them back to this thread when I get a chance. Clearly audible should be easily measured.

I'm not keen to go out and buy two DACs to test this myself. I'm not made of money unfortunately! Do you have access to a way of transparently capturing the outputs of the DACs to a digital file? That would be the easiest way to distribute the files to all of us so we can see and hear exactly what we are talking about. Maybe even set up a null test to isolate the difference clearly. Or perhaps someone here with testing equipment will be able to verify specs on one of your DACs that is loaned for testing.

There's an order to what I'm doing here...

1) observation
2) verification
3) quantification / measurement
4) hypothesis
5) testing hypothesis

We are on step 2. We shouldn't jump ahead of ourselves. Let's think of a way to verify what you heard. It's clearly audible, so that shouldn't be difficult. Does that make sense?

By the way, I didn't see an answer to this question.... Which one of these DACs do you suspect is transparent, and which one do you think is colored? Have you compared either of them to other DACs and found them to sound the same? Obviously one has to be colored, because if both were transparent, they would both sound the same.

 

[KeithEmo]

Personally I always preferred the Nikon d800 version without the anti-aliasing filter

Keith...

The "slightly blurred negative image" = lowpassed image with inverted phase

Optically summing them = cutting all frequencies except those cut by the lowpass

Further lighting up the summation = boosting the frequency bands cut by the lowpass.

And no, "an actual halo" is not the preferred result. If so why don't you add some actual halos to your perfectly sharp LCD monitor using the sharpness control that may or may not be there anymore?



The effect of the interpolation filter is always there.
1. If there's no interpolation filter you get no sound.
2. If a non-oversampled zeroth order hold is used, high frequencies pure tones will sound very funky indeed (like a square wave of the frequency f modulated by a Nyquist-2f beat frequency, AFAICT). But a transient signal won't ring!
3. If a proper filter is introduced, it is close to as you say, with respect to their analog originals. With the exception of the last point: within the confines of sampling theory, there's no such thing as a transient signal that ought to appear as input to a DAC. A sine wave start starts and stops on a dime, or a Kronecker delta, are artificial constructs that can never appear in a real-life recording, because they are cropped versions of theoretical signals that contain frequencies extending beyond Nyquist out to infinity. The "introduced ringing" is simply what happens when these theoretical signals are brought back to reality, rendered using the finite set of frequencies available under the sampling format.

Speaking of which, there IS a theoretically perfect reconstruction filter against which all real life reconstruction filters are to be measured: the infinite length sinc filter. It will produce ringing tails before and after your sine burst and K-delta--infinitely long tails! But said ringing would only consist of frequencies a fraction of a Hz under 22.05kHz (for 44.1kHz sample rate) and hence be completely inaudible (to you and me anyway)!

As noted in a previous post--the *shorter* the ringing response to a K-delta, the more artifacted the filter's response to real life input--including music and high frequency sine tones!



As noted before, the ONE WAY in which I could detect imperfect filtering in pedestrian DACs was by listening for aliasing in high frequency sine tones--NOT by listening to K-delta clicks, square waves, theoretical sine pulses, or anything like that!



There's an antialiasing filter in front of the sensor to prevent that scenario and to replace that with, exactly, "a drop in high frequency response". If not, the sensor is high resolution enough that lens aberrations take care of the antialiasing. If not, serious artifacts occur when shooting finely patterned fabrics.

--Joe Bloggs
--10+ year dpreview member
--designed sharpening-based filter for HDR toning that sharpened *everywhere in the image except where halos would occur*: https://www.dpreview.com/forums/thread/2902464#forum-post-36868711

 

[castleofargh]

"We used to have one around here
that the minimum acceptable dynamic
range for recorded music is 120dB."

And who might that have been?

amirm
he's not an uberman believer like some are. he simply defines audibility with a wider range of conditions than we do. including loud output, quiet passages, very isolating IEMs... so obviously the thresholds found under such circumstances are going to be different from the thresholds found with music averaged around 70 or 80 dB SPL, and more typical albums that will rarely show more than 50 or 60 dB down for the quiet parts.
basically he says that if some conditions make something audible, then it is. which isn't wrong, but tends to take us pretty far away from what anybody would get with typical music listening. which is admittedly still what most of us care about.

 

[KeithEmo]

And which of the twenty or so iPod models did you compare to which of the hundreds or thousands of devices that use the rather popular Sabre DAC chip?
And, by the way, which speakers or headphones, and which source content, did you use to compare them?

Obviously it would be extremely un-scientific to try and generalize one particular pair of samples to "all DACs everywhere".

If one comparison, under one specific set of circumstances, proves an audible difference, then we must concede that "audible differences sometimes exist".
However, we must cover a lot more ground to "prove" that audible differences "never" exist (or even "occur rarely enough that we can ignore them").
For starters, did the device you compared, with the Sabre DAC, use the internal filters provided by ESS (Sabre), or did the designers use their own external filters?
Most folks I know who support the idea that differences exist also find that "some Sabre DACs have more of the characteristic Sabre sound than others"
(But, as I mentioned in another post, that Sabre DACs OFTEN sound different than other brands, and that, when they do, the difference is virtually always of the same character.)
It has also been noted, by some of the people who imagine they hear a difference, that Sabre DACs are more sensitive to the circuitry that accompanies them than most others.

The obvious way to confirm the basic claim would be to compare two DACs that one or more people insist "are obviously different in sound"...
And see if that difference is confirmed or denied in a proper test.
Assuming that the difference is really obvious, that should be possible with two DACs, a few people, and some good samples and associated equipment.
(But, no, I see no way to avoid actually placing the two DAcs in front of the people performing the test.)

Why? I'm not trying to determine all of the differences, just the audible ones. If a capture device is capable of audible transparency, then it should be fine for reproducing audible differences.

There's no point measuring inaudible differences, and I'm sure they exist. I am trying to find a way to verify that two DACs sound different to human ears. If we're assuming there must be a difference because one is a Sabre chip and one isn't, I have a headphone amp with a high end Sabre chip and it sounds identical to an iPod, so it isn't that.

That is the core problem; you always insist that any difference in > 20 kHz, lower than so and so % THD, etc, performance has no effect on audible differences. If you can, try to have both micros side by side - an afternoon is more than it takes, regardless how you will compare them. It is the only case of a commercialy available device known to me where significantly better electronic parts - when taken together - clearly result in an audible improvement. The THD, etc numbers are crazy in any case - and are particularly off the charts with Black Label. It should not matter audibly - yet, it does. Because the SQ is largely determined by analogue stages - input in ADC, output in DAC.

It might answer the question " How transparent is transparent? " ...

 

[Jaywalk3r]

I will, however, note that I have had a lot of conversations with a lot of people about "how DACs sound" (since part of my job is providing support for customers who purchase DACs).
In find that roughly somewhat more than half of the people I talk to, if asked, agree that Sabre DACs "seem to have a distinctive sound".
Furthermore, OF THOSE WHO CLAIM TO HEAR A DIFFERENCE, almost all of them describe the difference they think they hear quite similarly, whether they like or dislike it.
(Those who like it say that they find Sabre DACs to be "more detailed"; those who dislike it consistently describe Sabre DACs as sounding "more grainy" or "etched".)
Feel free to consider this to be "a sociaological study" if you like... but the majority of people who believe they hear a difference seem to believe they hear a SIMILAR difference.
(And, oddly, it is much the same difference that I think I hear.)

Again, you've described a Petri dish for cognitive biases.

there is very little "pure science" left in the world - and virtually none when it comes to audio equipment.

There's very little science in high-end audio; it's primarily snake-oil, at least outside of speakers. The use of science is, in general, increasing in the world.

Speaking of science, if you claim there is a difference in DACs, the onus is on you to support that claim with evidence. If you don't do that, reasonable people are fully justified to dismiss your claims out of hand, no matter your background. That's science.

A company who sells $500 interconnects has little incentive to do a proper study (because it will probably show that their product is eitehr snake oil or only marginally better).

Right. Demonstrating to people that they are wasting their money if they buy your product instead of a much less expensive alternative is generally a bad business strategy. On the other hand, a manufacturer who is selling something that isn't snake-oil has every incentive to demonstrate that their product provides value. If all they provide are excuses, it's a safe bet they are selling snake oil.

Yet the company who sells $5 interconnects ALSO has little incentive to sponsor the test.

Correct. They aren't making any ridiculous claims about their interconnects, other than they will connect one component to another, which is easy to demonstrate conclusively.

What the manufacturer of those $5 interconnect is doing, in most cases, is methodically sampling and testing production units so that reliable estimates about production yield rates can be made with known levels of confidence, using well-established statistical methods.