Serious Question: How can DAC's have a SOUND SIGNATURE if they measure as transparent? Are that many confused?

[solderdude]

In general the original question by the autor was not responded concisely: I let a resume of a few notes are taken to the question “how can DACs measured transparent can sound different?”

-They don’t (majority)

requires a definition of measured transparent.

-Filters alters the sound (some members)

Can easily be measured. Consider that slow filters do not measure transparent so there's that ... and the age of the listener as well as the sample-rate.

-Measurements may fail to predict transparency (my suggestion that caused a lot of reactions, and probably audiophile believes)

Depends on what is measured and how various measurements are combined to determine technical performance.
Problems occur when trying to predict sound quality and subjective found differences due to the huge amount of possible variances opposite lab condition measurements.

It is this part that is questioned... your personal experiences in your situation.
I would suggest you stop buying ifi DACs... they do not seem to be your thing. We (as well as you) don't know the reasons for it.

-Other unknown variables (again my suggestion, same as measuring incompleteness but differently formulated)

Good thing there is such a thing as 'nulling' where one can even listen to nulls.

 

[Basic Channel]

On the thread nobody quoted a single concrete study answering the question posted. Many members talk about “30 years of experience” but not supported by any publications.
Can you give me at least one or even more links to those studies?

On this thread you have repeatedly stated that you find a particular DAC fatiguing, have shown no evidence to explain why it might be other, and attempted to blind side people with waffle about statistics.

You have additionally said mp3 320kpbs is lossless, and that you had difficulty differentiating that from 128kbps which is a far greater difference than any non broken DACs.

Why should anyone provide you with evidence when they could just write some long prose saying it is like the Keynesian liquidity trap and talking about “physics envy”.

 

[olieb]

Until then there's nothing to research.

To make the point even clearer. There is a genuine interest in exploring the audibility threshold for distortion (harmonic, intermodulation, noise,...). But this has already been done (i.e. in the context of lossy codecs and more generally).
The thresholds are so much higher than what is measured with "transparent" DACs that the onus is now on the people who claim they still hear differences (whether because of golden ears, "unmeasured" new forms of distortion/effects, or whatever) to provide controlled proof.

Otherwise it is the same as claiming there is a large teapot inside the moon and pointing out there are no scientific papers proving the opposite.

 

[ads_cft222]

With "the question posted" you mean "how can DACs measured transparent can sound different”? If so, you shouldn't expect people who don't believe audible differences exists to come up with publications. That's Russell's teapot whole over again. It's the believers that need to come up with evidence. Until then there's nothing to research.

I think answers as "No difference at all" and "Big difference" are both extremes and both need verification.

Why don't you all settle to the statement: Speakers and room (Treatment eq) and subs are the most important, then amplifiers and last is dac. As long as someone spends sensible money with the correct priorities he is making sensible choices.

 

[Miguelón]

With "the question posted" you mean "how can DACs measured transparent can sound different”? If so, you shouldn't expect people who don't believe audible differences exists to come up with publications. That's Russell's teapot whole over again. It's the believers that need to come up with evidence. Until then there's nothing to research.

“Measured transparent” means “linear FQ response”, “under threshold distortion” and “under threshold noise”?
Is what I understand, there are electronic tests, not audition tests.

I submit the question to our ORL who conducted many trials about audition.

His answer, resumed:

“All auditions test should be taken under consistent variables. Actual auditive measurement where validated under decades of research including stress factors, room acoustics, ethnical variations, age, gender, underlying pathologies and other factors.

Many trials reviews show that at least 50% of audition clinics or physiology failed on methodology: inconsistent results should not be attributed to patients bias before carefully reviewing the methods. Principal factors that were attributed to altered results were stress, pre noise exposure before testing, unacceptable biased recruitment and inadequate environment

I don’t know about physical audio gear testing, nor speakers nor dacs, but at the actual date and given that hearing human physiology is not fully understood we still need better setups even at high clinical level research to validate acoustic measurements. So I don’t think anyone in other field can definitively confirm many of our beliefs, we should very prudent on each statement.

As an example, many of my patients are stressed when worrying about possible adverse results from audiometry. If that can distract from single sinusoid frequencies when asking the patient to click the button when hearing, I cannot imagine how difficult will be to measure to music subtle differences”

When I mentioned Russel teapot regarding to non-objetive perception he answer “welcome to sensorial world, many patients suffer from tinnitus and nobody has developed a method to prove their existence.

Conversely, some patients show an audition threshold of -15 dB, others came from +60 dB or more. Prior to measure some subtle differences on some acoustic phenomenon perhaps you should select different types of population to determine weather the phenomenon exist or not”

 

[Miguelón]

On this thread you have repeatedly stated that you find a particular DAC fatiguing, have shown no evidence to explain why it might be other, and attempted to blind side people with waffle about statistics.

You have additionally said mp3 320kpbs is lossless, and that you had difficulty differentiating that from 128kbps which is a far greater difference than any non broken DACs.

Why should anyone provide you with evidence when they could just write some long prose saying it is like the Keynesian liquidity trap and talking about “physics envy”.

I never said 320 kbps is loseless, please read more carefully. Loseless audio on the test I made were provided at CD quality, there were 3 tracks (same original) on different resolutions: WAV uncompressed, 128 mp3 and 320 mp3. The question was to guess which one has the best audio quality in the subject perception.

I don’t know what is the definition of loseless, from the term is often used that means uncompressed files from 16/44.1 or above, but perhaps I’m wrong

 

[Geert]

I think answers as "No difference at all" and "Big difference" are both extremes and both need verification.

To take into account it's not about "No difference at all', it's about measurable and audible differences. We know there are differences, and we can measure them. So in the end the discussion boils down again to the believe that there are audible differences we can't measure, including not show up in null tests.

 

[Purité Audio]

Comparing two dacs unsighted and level matched really isn’t that stressful.
Keith

 

[Basic Channel]

I never said 320 kbps is loseless, please read more carefully. Loseless audio on the test I made were provided at CD quality, there were 3 tracks (same original) on different resolutions: WAV uncompressed, 128 mp3 and 320 mp3. The question was to guess which one has the best audio quality in the subject perception.

I don’t know what is the definition of loseless, from the term is often used that means uncompressed files from 16/44.1 or above, but perhaps I’m wrong

You are correct I misread your post. Fortunately we don’t fully understand human psychology or the origin of language so we can’t actually say I was wrong. Maybe I was actually correct but we just don’t understand why yet. In fact, I’ll waffle here about Chomsky and grammar and the wanna contraction.

 

[Geert]

welcome to sensorial world, many patients suffer from tinnitus and nobody has developed a method to prove their existence.

Bad analogy since tinnitus is not real sound, instead it's again the brain playing tricks. So a tinnitus patient should not ask healthy people to go find and disable that disturbing noise, instead we should analyze the patient's brain. Which has been done successfully Tinnitus and distress: an electroencephalography classification study ("The combination of EEG with sophisticated classification procedures may reveal biomarkers that can identify tinnitus and accurately differentiate different levels of distress experienced by patients").

 

[Miguelón]

Bad analogy since tinnitus is not real sound, instead it's again the brain playing tricks. So a tinnitus patient should not ask healthy people to go find and disable that disturbing noise, instead we should analyze the patient's brain. Which has been done successfully Tinnitus and distress: an electroencephalography classification study ("The combination of EEG with sophisticated classification procedures may reveal biomarkers that can identify tinnitus and accurately differentiate different levels of distress experienced by patients").

Yes, but never tried patients to prove the existence of tinnitus. I know the article but is not necessary to make a MRI to diagnose tinnitus: we trust our patients

 

[antcollinet]

Yes, but never tried patients to prove the existence of tinnitus. I know the article but is not necessary to make a MRI to diagnose tinnitus: we trust our patients

Why should they need to? No one here is arguing with the fact that you perceive a sound difference, just as no-one argues that tinnitus sufferers hear their tinnitus.

What is argued (in audio, not in tinnitus) is where those perceptions come from. When you perceive differences between devices that measure as transparent (imperfections below the level of audibility), then it is clear those perceptions don't come from the sound waves. Perceptual biases must be at play.

 

[Miguelón]

Why should they need to? No one here is arguing with the fact that you perceive a sound difference, just as no-one argues that tinnitus sufferers hear their tinnitus.

What is argued (in audio, not in tinnitus) is where those perceptions come from. When you perceive differences between devices that measure as transparent (imperfections below the level of audibility), then it is clear those perceptions don't come from the sound waves. Perceptual biases must be at play.

Also measuring biases exist. Tests that are used to measure dac are just send a signal and analyze the response. After that, you believe that the dac will behave consistently over a whole track of 5 minutes (as an example) and with quite superposition of waves. I guess they are other tests more complex as multitone.

But the only way to state that two different dacs are equal on a given track is to analyze the whole function of the output signal. The other are samples.

I suppose that the device is well known and wave superposition do not overcome the circuit with interferences, overheating or any other phenomenon and this idea can be very weak, but think that it will be a better measurement than single samples.

What is very surprising to me is given the fact that discrepancies exist between measurements and people perception, and also blind tests cannot be done with enough accuracy, nobody take the possibility of inconsistency of the method. Always attribution to psychological phenomena is not the unique way to explain.

I’m not economically concerned with the dac topic, but if you want really to definitively state dac equality why to not perform a complete signal analysis?

One way may be reconverting signal output to digital with an ADC, on two different dacs (but same ADC of course), and submit the resulting files to Fourier analysis or even AI search for looking different patterns.

Or maybe this has be done, I don’t know

 

[undeakyetonde]

Tests that are used to measure dac are just send a signal and analyze the response. After that, you believe that the dac will behave consistently over a whole track of 5 minutes (as an example) and with quite superposition of waves

https://en.wikipedia.org/wiki/Linear_time-invariant_system

 

[Jim Taylor]

What is very surprising to me is given the fact that discrepancies exist between measurements and people perception,

Why is this surprising? Measurements are controlled, dispassionate scientific information. Perception is subjective opinion controlled by the mind's biases. To me, there is no surprise at all.

Perception is not calibrated; it does not exist in relation to a verifiable standard. Measurements do.

Jim

 

[antcollinet]

After that, you believe that the dac will behave consistently over a whole track of 5 minutes (as an example) and with quite superposition of waves

Yep - because that (within limits) is how electronics works.

I guess they are other tests more complex as multitone.

Indeed - but really not needed for an understanding of performance. It is just there to answer the "but you are only testing with test tones"

What is very surprising to me is given the fact that discrepancies exist between measurements and people perception

No that is not surprising. We know that peoples perceptions (of all types) are influenced by biases (of all types). Measurements are not.

and also blind tests cannot be done with enough accuracy,

Who says this - you can blind test yourself fairly easily with a small investment of time and equipment. If as you say the differences are easy to hear, you should easily pass the blind test as long as the difference is actually in the sound waves.

nobody take the possibility of inconsistency of the method

What method - the test method used here? Absolutely consistent, done with the same (incredibly accurate) test gear, to the same test plan by a very knowledgable engineer - time after time.

Or maybe this has be done, I don’t know

It has been - one of our members has even created some software to easily compare digital files. Search here for "deltawave"


EDIT:
Here is a post where two DACs were compared using Deltawave:

Measuring the "sound signature" of two different digital sources (DSD 2.8MHz)

Measuring the "sound signature" of two different digital sources (DSD 2.8MHz) https://www.audiosciencereview.com/forum/index.php?threads/marantz-sa-11s2-review-sacd-player-seiko-npc-sm5866as-monaural-dac.37986/ Vs...

www.audiosciencereview.com

I'm not sure if I should be dropping the mic here or not.

 

[DLS79]

I've often wondered if the people who post utter ridiculousness here are just the same handful of people cycling through new accounts!

 

[IAtaman]

What is very surprising to me is given the fact that discrepancies exist between measurements and people perception, and also blind tests cannot be done with enough accuracy, nobody take the possibility of inconsistency of the method.

That is not surprising at all. I'd wager at any point no more than 50% of our opinions would align with physical facts, and the rest will be idiosyncrasies of our existence and culture.

 

[IAtaman]

I've often wondered if the people who post utter ridiculousness here are just the same handful of people cycling through new accounts!

Audiophiles would be the easiest and least interesting social group to automate with AI indeed.

 

[Blumlein 88]

Also measuring biases exist. Tests that are used to measure dac are just send a signal and analyze the response. After that, you believe that the dac will behave consistently over a whole track of 5 minutes (as an example) and with quite superposition of waves. I guess they are other tests more complex as multitone.

But the only way to state that two different dacs are equal on a given track is to analyze the whole function of the output signal. The other are samples.

I suppose that the device is well known and wave superposition do not overcome the circuit with interferences, overheating or any other phenomenon and this idea can be very weak, but think that it will be a better measurement than single samples.

What is very surprising to me is given the fact that discrepancies exist between measurements and people perception, and also blind tests cannot be done with enough accuracy, nobody take the possibility of inconsistency of the method. Always attribution to psychological phenomena is not the unique way to explain.

I’m not economically concerned with the dac topic, but if you want really to definitively state dac equality why to not perform a complete signal analysis?

One way may be reconverting signal output to digital with an ADC, on two different dacs (but same ADC of course), and submit the resulting files to Fourier analysis or even AI search for looking different patterns.

Or maybe this has be done, I don’t know

Everything in this post is wrong and/or unaware of existing tests that do what you are supposing is not done. The fact you don't know this is okay. The fact you refuse to learn when we tell you otherwise is not. The statement you made in another post about having no doubt in what you hear is anti-scientific.

You can search the forum for Deltawave and Multitone and pkane. User pkane has software that covers all of your ideas about testing dacs.