[bigshot]

"As a chef, I don't see much value in preparing food".

I think that lets us know he isn't anything even remotely resembling a scientist.

Does Apple USB/amp reproduce all the frequencies equally not rolling off 15-20 kHz for instance? Do we have a clear proof of this?
I am open to correcting my misconcepts based on the tangible evidence.

https://www.kenrockwell.com/apple/lightning-adapter-audio-quality.htm
Quote: "It's flatter than most audio analyzers. I measure -0.075 dB at 18,500 Hz, -0.15 dB at 19,250 Hz, -3.5 dB at 20,900 Hz and then it cuts off sharply to -95 at 21,700 Hz. This is flawless; bravo!"

That's a review of an $8 DAC. Here is a review of an older iPhone's headphone output.

https://www.kenrockwell.com/apple/iphone-6s-plus-audio-quality.htm
"ruler flat driving real loads"

 

[gregorio]

As a scientist, I do not see much value in controlled tests for DACs here specifically, it may only prove that I can hear the difference (not universal) and here it will be just more sceptical questions from guys like bigshot.

Hmmm. As a scientist you must presumably be aware of basic science? For example, the basic principle of the null hypothesis? In audio, the common null hypothesis is that there is no audible difference. If you can hear a difference you have therefore falsified the null hypothesis! Of course though, you do actually have to hear an audible difference, not just perceive a difference due to some bias/perceptual error and that means a well controlled test. Furthermore, if you do a controlled test and falsify the null hypothesis, be prepared to have your methodology scrutinised/criticised if your result is contrary to the results obtained by others who’ve done the same controlled testing. But as a scientist I don’t need to tell you that because you already know that experimental results must be reproducible by others to even start to be considered reliable evidence.

I have a bit hard time to understand why it is not reasonable to assume that some DSP is inevitably applied to most amps and daps in audiophile products?

Because most amps are analogue devices which therefore do not perform any DSP. Most Daps on the other hand obviously include DACs and almost all DACs do necessarily perform DSP. The question though isn’t if there is any DSP occurring (because there almost always will be), the question is: Are the effects of that DSP above the threshold of audibility?

Does Apple USB/amp reproduce all the frequencies equally not rolling off 15-20 kHz for instance? Do we have a clear proof of this?

Firstly, as the question is as just stated above, why are you asking this? You’ve picked the audio region (15-20kHz) with the highest audibility threshold. In fact, much/most of that region has an audibility threshold higher than is possible at reasonable listening levels for adults beyond their late teens. IE. It doesn’t matter whether we’re rolling off 1dB or 60dB it won’t be audible (for adults in much of that range).

Secondly, we do have reliable evidence and I’ve already presented it to you! Quoting the first link:

“I measure -0.075 dB at 18,500 Hz, -0.15 dB at 19,250 Hz, -3.5 dB at 20,900 Hz and then it cuts off sharply to -95 at 21,700 Hz.”

Again, not even close to audible. -0.15dB might be just barely detectable with a highly trained, far more sensitive than average listener with perfect hearing, if it were a specifically designed test signal in the most sensitive hearing region (around 3kHz) but there’s not even a remote chance at 19kHz with a music signal!

However, there are a few pathological cases of DACs with larger roll-offs far lower in the audio spectrum, which can be audible under certain conditions; Some NOS DACs and some filter choices to emulate such NOS DACs for example.

G

 

[Corti]

That is not even an issue. The analogy doesn't make sense for other reasons.

A DAC has to convert a given digital input signal to an analog output signal.
It is possible to do this with enough precision such that (amplified and played back over transducers) it can not be audibly distinguished from an even more precisely - or even perfectly - converted signal (amplified and played back over the same transducers). Which means it can be done audibly transparent or audibly perfect.
What now determines what pleasure you get from it, is the content of the digital signal, the music or movie sound or whatever, that should be according to your taste.
This is a compeletely different situation from tasting wines. The wines themselves are the content. They are not supposed to "perfectly convert" some input coming from somewhere else. So there is no concept of transparency. Different wines taste different like different music pieces sound different. If you want to compare a DAC with something wine related, I'd sooner compare it to a glass to drink the wine from (although that also isn't a really good analogy).

Good point, but i still want that complete list of wine molecules (including structural geometry). maybe i can then 3D-print my own tastey-ophile wine

 

[hakunamakaka]

How some of you is not bored arguing this stuff over and over again is beyond my comprehension. Do you actually have time to listen to music ? There is more to life than head-fi

 

[james444]

But you weren’t just making an empirical observation that you got more goosebumps, you also “judged” what was causing it (euphony).

If you read my OP carefully, I actually didn't. It was my objectivist friend who came up with the term. Anyway, that's just a side note and not really important.

Maybe the difference with your ESS based DAC is the filter option used by its designer, (if it’s say the Sabre 9018) some manufacturers allow user switching between various filter options programmable by software on the chip itself, “slow roll off” akin to NOS maybe the one chosen for the “analoguish “ sound,
But that then begs the question, if another manufacturer uses a different filter option that sounds slightly different which option would be considered “audibly transparent” ?
Why, according to some, would a DAC chip manufacturer allow an option to switch between the “audibly transparent” option and three or more “broken” ones ?
Or maybe it’s just a matter of fine tuning to suit various output stage differences ?

I have an LG V30 with Sabre 9218 DAC. This mainstream chip can be found in many portable devices, like smartphones, DAPs and USB-dongles. According to the manufacturer ESS it offers a "fully programmable FIR filter with eight presets provides a customizable sound signature". On my LG V30, three of these presets are selectable via user interface and they do indeed sound slightly different.

IMHO, this makes the whole discussion about all DACs being audibly transparent somewhat moot, because it's obviously very unlikely that all implementations of this DAC chip will use the same FIR filter configuration. Not that it really matters in the grand scheme of things, since the actual sound differences should be tiny. However, as you rightly stated, only one out of all possible configurations could be considered "audibly transparent".

As for your last two questions, I found the following info on the three DAC filter presets on my V30. They're named short, sharp and slow.

Short I think is what is generally called minimum phase. It removes preringing on the impulse response, but causes phase shift across frequency response.

Sharp is the normal setting on most things.

Slow gives less ringing on the impulse response but there will be more aliasing.

According to where I got the info from, it seems to be related to a "never-ending debate of linear versus minimum phase filters":
https://troll-audio.com/articles/linear-and-minimum-phase/

That's all I can say on this topic, as I neither know much about DACs, nor do I care much about tiny differences in their sound signatures.

 

[PhonoPhi]

Hmm, ok can you give me a list of all the molecules in a certain wine? If you say that it is possible, then there must be studies/papers about it (or is it top secret?). How else do you know about the possibility?

My knowledge is that we are far from knowing all the exact ingredients (requiring the chemical isolation of every single different molecule) within different plants and their fruits. We can certainly do mass spectrometry or chromatography, but that doesn’t tell us all about their atomic-structure (which is critical to know when we want to know how they react with taste-buds and the olfactory system).

What i could think of is that we can measure parts of the ingredients (those that we already chemically isolated) and link them to the five human tastes and say: this wine contains a certain amount of x alkaloids, so that the wine probably tastes y amounts sweet, z amounts salty etc. but its only an approach and far from being exact.

While we can measure the output of amps/dacs very precise. and we are talking about identical and not about similar.
But if that’s not the case and we can measure exactly every single molecule I would seriously like to read about it.

There are limited classes of molecules.
So true, identifying every single molecule is not feasible, while identifying anything required for specific purpose is a reality these days stemming from the advances in biochemical analysis.
Here is recent relevant review on approaches to identify the composition: https://link.springer.com/article/10.1007/s12161-021-02138-6
The lower the concentration of molecules - the more concentration techniques needed - the subject of this review.

I am not sure what you would print "3-D" but making an artificial wine won't be much of a problem to 99+%
Water, ethanol, sugars, carbon dioxide, organic acids, phenolics, esters, yeast fragments - would be sufficient.

Proving to be a scientists here would beeven more ironic than doing controls for bigshot...

 

[PhonoPhi]

"As a chef, I don't see much value in preparing food".

I think that lets us know he isn't anything even remotely resembling a scientist.


https://www.kenrockwell.com/apple/lightning-adapter-audio-quality.htm
Quote: "It's flatter than most audio analyzers. I measure -0.075 dB at 18,500 Hz, -0.15 dB at 19,250 Hz, -3.5 dB at 20,900 Hz and then it cuts off sharply to -95 at 21,700 Hz. This is flawless; bravo!"

That's a review of an $8 DAC. Here is a review of an older iPhone's headphone output.

https://www.kenrockwell.com/apple/iphone-6s-plus-audio-quality.htm
"ruler flat driving real loads"

Would the response hold when multiple frequencies are applied (as in real.music).
Would they be equally represented?
Do you have any evidence for it?
Frequency sweeps are useful to the extent, their limitations are obvious in transducer characterization.

 

[bigshot]

If the filter isn't user configurable, I'm sure the manufacturer is going to set it to use the default (read: transparent) filter.

 

[gregorio]

it's obviously very unlikely that all implementations of this DAC chip will use the same FIR filter configuration. Not that it really matters in the grand scheme of things, since the actual sound differences should be tiny. However, as you rightly stated, only one out of all possible configurations could be considered "audibly transparent".

Why is it obviously very unlikely they will use the same filter configuration? Switchable filters have been around for quite a few years and in all cases I’ve seen, the standard, fast filter is the default.

Also, typically most of the filter presets are audibly transparent, typically only one isn’t, the slowest, which often rolls-off starting around 10kHz and so will be audible with some music, assuming decent hearing up to at least 12kHz-14kHz. However, specs I’ve seen for other chips had 5 switchable filters so with 8, the chip you’ve quoted might have more than one that’s potentially audible.

Different DACs have typically always had somewhat different filters (though not switchable). Many had filters starting around 20kHz, some at 19kHz, some 18kHz or even 17kHz or so and some slightly higher than 20kHz. They are all different sound signatures and they are all audibly the same! Now you can get all those inaudibly different filters in one DAC, plus maybe one or two dodgy filters you might be able to actually hear. It’s pretty much all nonsense but great for audiophile marketing, which is why they’ve done it of course!

Would the response hold when multiple frequencies are applied (as in real.music).

No, it would be even more inaudible! In music the majority of the energy is in the lower frequency range, so you’ve almost certainly got significantly lower levels of very high freq content to start with. AND, with a real/music signal and multiple frequencies, there’s a high probability of audio masking occurring in that very high freq range, even assuming the freqs above 17kHz-18kHz were potentially audible in isolation!

G

 

[Corti]

Here is recent relevant review on approaches to identify the composition: https://link.springer.com/article/10.1007/s12161-021-02138-6

So I was able to download the paper because I have an account that lets me access springer scientific content for free instead of paying 40€.
In the paper they talk about extracting substance classes and how they contribute to the quality of the wine.
But they do not at all measure anything like the taste of the wine. They just show different extraction methods. Search function did not even find the term ‘taste’

But I didn’t expect otherwise.. so the comparison of wine taste and sound of ‘audiophile products’ is still non-sense to me.

Water, ethanol, sugars, carbon dioxide, organic acids, phenolics, esters, yeast fragments - would be sufficient.

Lol so there is only 1 sugar, 1 organic acid, 1 phenolic, 1 ester and 1 yeast fragment in the world, no variations.. I see, then of course it should be easy to make artificial wine, perfect manual.. i'll just go to supermarket and buy 1 Water, 1 ethanol, 1 sugars, 1 carbon dioxide, 1 organic acids, 1 phenolics, 1 esters and 1 yeast fragments

I am not sure what you would print "3-D" but making an artificial wine won't be much of a problem to 99+%

“not much of a problem to 99+%,” whatever that is supposed to mean.
But you also didn’t understand that the 3-d printing was a joke..
I guess I’m wasting my time talking about nonsense which does not relate to what was discussed before.

also your quote from earlier

We can measure about every molecule in wine and link it to taste

is still unproven. we can not measure about every molecule in wine. we can only extract certain classes of substances.

But this has already been highly off-topic.

 

[bigshot]

Would the response hold when multiple frequencies are applied (as in real.music).
Would they be equally represented?
Do you have any evidence for it?
Frequency sweeps are useful to the extent, their limitations are obvious in transducer characterization.

In DACs, the highest frequencies up by the cutoff are the ones that are most likely to have deviation from a flat response. If you notice, the spec becomes a bit worse, the higher you go... Rockwell says, "I measure -0.075 dB at 18,500 Hz, -0.15 dB at 19,250 Hz, -3.5 dB at 20,900 Hz and then it cuts off sharply to -95 at 21,700 Hz." If it is within .075 dB at 18.5kHz, you can be sure it's as balanced as it can possibly be a few octaves down. Likewise, if the response is balanced with tones far beyond the threshold of perception, with music, it's going to be even further below the the threshold. Human ears are much less forgiving of deviations in tones than they are in music. The measurements I linked you to show that the Apple dongle is audibly transparent by more than one order of magnitude. The specs are significantly better than the industry average.

I am open to correcting my misconcepts based on the tangible evidence.

I think I'd like to see some evidence to back up this comment right about now. You've received the tangible evidence of your "misconcepts".

 

[PhonoPhi]

So I was able to download the paper because I have an account that lets me access springer scientific content for free instead of paying 40€.
In the paper they talk about extracting substance classes and how they contribute to the quality of the wine.
But they do not at all measure anything like the taste of the wine. They just show different extraction methods. Search function did not even find the term ‘taste’

But I didn’t expect otherwise.. so the comparison of wine taste and sound of ‘audiophile products’ is still non-sense to me.


Lol so there is only 1 sugar, 1 organic acid, 1 phenolic, 1 ester and 1 yeast fragment in the world, no variations.. I see, then of course it should be easy to make artificial wine, perfect manual.. i'll just go to supermarket and buy 1 Water, 1 ethanol, 1 sugars, 1 carbon dioxide, 1 organic acids, 1 phenolics, 1 esters and 1 yeast fragments


“not much of a problem to 99+%,” whatever that is supposed to mean.
But you also didn’t understand that the 3-d printing was a joke..
I guess I’m wasting my time talking about nonsense which does not relate to what was discussed before.

also your quote from earlier

is still unproven. we can not measure about every molecule in wine. we can only extract certain classes of substances.

But this has already been highly off-topic.

Off- topic it is; the main components of the wine taste and smell were established quite a long time ago.
Great that you could access the review - I selected it as an example that all components can be concetrated, isolated and measured. The intro gives a good brief overview of other methods.
Looking through the reviews, I was a bit surprised that the artificial nose sensors are not used more. Traditions, and then what is needed is studied, but not much else, it is more applied field than pure science.

The concentrations of main components are known, surely different combinations are infinite, but to make a decent synthetic wine undistinguishable by taste at least to a table wine should not be any problem.

Similarly, knowing what to measure with DAC/amps would be much more instructive that general talk about "transparency".

 

[PhonoPhi]

In DACs, the highest frequencies up by the cutoff are the ones that are most likely to have deviation from a flat response. If you notice, the spec becomes a bit worse, the higher you go... Rockwell says, "I measure -0.075 dB at 18,500 Hz, -0.15 dB at 19,250 Hz, -3.5 dB at 20,900 Hz and then it cuts off sharply to -95 at 21,700 Hz." If it is within .075 dB at 18.5kHz, you can be sure it's as balanced as it can possibly be a few octaves down. Likewise, if the response is balanced with tones far beyond the threshold of perception, with music, it's going to be even further below the the threshold. Human ears are much less forgiving of deviations in tones than they are in music. The measurements I linked you to show that the Apple dongle is audibly transparent by more than one order of magnitude. The specs are significantly better than the industry average.


I think I'd like to see some evidence to back up this comment right about now. You've received the tangible evidence of your "misconcepts".

I am fine with my "misconcepts" and wish you the same.

For the specs, to start with - 37.5 Ohm is not totally unrealistic, but far from typical 16 Ohm.

For the impedance, I have already seen three values - 0.3 Ohm, 0.9 Ohm and 1 Ohm (from what you cite).
0.3 Ohm is not likely to be right, but that is still the same type of enthusiatic measurements googled and cited...

 

[Dogmatrix]

That's it? Most 'dissertation' titles tell the potential reader something about the original research conducted.

That was it , within the context of the specific degree it was considered sufficient although I was advised of some risk

 

[james444]

Why is it obviously very unlikely they will use the same filter configuration? Switchable filters have been around for quite a few years and in all cases I’ve seen, the standard, fast filter is the default.

How about "give people options and they'll choose"? How about probability and common sense?

These cheap ESS chips with configurable filters have been implemented by the hundreds, if not thousands in portable devices. They're all over sites like Alibaba in Chinese phones, DAPs and USB dongles. Sold by dozens of boutique audio companies that don't give a damn about what objectivists think. Yet you're absolutely convinced that all of these manufacturers use the same filter configuration? Really?

Footnote: I share your sentiments about audiophile marketing nonsense and dodginess. But this is strictly about the claim that all DACs sound the same, nothing else.

If the filter isn't user configurable, I'm sure the manufacturer is going to set it to use the default (read: transparent) filter.

So you're "sure". But that isn't exactly science, is it?

Sorry about being somewhat provokative. But you wouldn't let a post like yours pass, if it came from someone else. Just applying your standards to yourself here.