Head-Fi.org › Forums › Equipment Forums › Sound Science › The Hopelessly Derailed ODAC/Objective DAC Anticipation/Discussion Thread
New Posts  All Forums:Forum Nav:

The Hopelessly Derailed ODAC/Objective DAC Anticipation/Discussion Thread - Page 2

post #16 of 256

?

Quote:
Originally Posted by kiteki View Post

 

Ah, thanks for the correction!

 

 

If I fail my future blind-test between my DAC and the ODAC (very likely), then I'll try another one with 24/192 upsampling versus 16/44.1 native.

 

The meyer & moran study was inherently flawed & statistically impossible according to this discussion, so I don't consider that survey valid for 24/192, DSD, or DAC transparency.

 

It may have tested A/D/A transparency, since one CD was the original, and the other went through an A/D/A loop, but both CD's were played back through the same equipment, in other words you may as well prove the transparency of mineral water and tap water drinking both mixed with Coke.

 

I haven't read it unfortunately - maybe I will sometime, but your analogy makes no sense. If the DAC was transparent how could it possibly matter?

 

Furthermore I would not trust the arguments of people who have wasted their money on a failed format. rolleyes.gif

 

post #17 of 256

 

Originally Posted by Satellite_6 View Post
Quote:
Originally Posted by kiteki View Post

in other words you may as well prove the transparency of mineral water and tap water drinking both mixed with Coke.

 

I haven't read it unfortunately - maybe I will sometime, but your analogy makes no sense. If the DAC was transparent how could it possibly matter?

 

If there are thousands of DAC chips, op-amps, capacitors, circuit layouts all with varying performance, filtering and inherent sound characteristics, then you can't assume that one of them is suddenly 100% transparent, for the sake of a test.

 

Example - If I take an original CD, and then create a duplicate CD which went through an ADC process, rip them both to my computer, put the files on a Hifiman HM-601, perform an extensive ABX - and can't tell the difference, what did I just prove?

 

 

post #18 of 256

 

Originally Posted by Satellite_6 View Post

 

Furthermore I would not trust the arguments of people who have wasted their money on a failed format. rolleyes.gif

 

Furthermore, I would not trust the arguments of people whom only ever look at one side of the coin.

 

On page 9 in that thread...

 

"David [Moran],

Thanks for the link. On page 777 of your paper you report:

"The “best” listener score, achieved one single time, was 8 for 10, still short of the desired 95% confidence level. There were two 7/10 results. All other
trial totals were worse than 70% correct."

Am I reading this correctly? There were approximately 50 subjects and only three managed to score 7/10 or higher? That is indeed a thin tail. Too thin. The probability of such a thin tail is something on the order of 0.0000000001.
"

 

See post 139 / 189 on page 14 for the follow-up.

 

post #19 of 256

 

Quote:
Originally Posted by kiteki View Post

 

 

Am I reading this correctly? There were approximately 50 subjects and only three managed to score 7/10 or higher? That is indeed a thin tail. Too thin. The probability of such a thin tail is something on the order of 0.0000000001."

 

 

 

 

Some of us are willing to take those odds if the payoff is worthwhile :  http://www.lotterywest.wa.gov.au/media/media-releases/archive/oz-lotto-leaps-to-50-million-jackpot

 

If I win, I'll be building a DBT studio dedicated to nothing beyond settling the endless arguments in the audio world. Wish me luck.  

post #20 of 256
Quote:
Originally Posted by estreeter View Post

Some of us are willing to take those odds if the payoff is worthwhile :  http://www.lotterywest.wa.gov.au/media/media-releases/archive/oz-lotto-leaps-to-50-million-jackpot

 

Well, it does not seem to be as unlikely as winning the lottery, I simulated 1000000 experiments with a simple program (assuming that all subjects are always randomly guessing), and the results are:

- 55 subjects, score_7 <= 2, score_8 <= 1, score_9 == 0, score_10 == 0:    4632 (0.4632 %)

- 50 subjects, score_7 <= 2, score_8 <= 1, score_9 == 0, score_10 == 0:    9195 (0.9195 %)

- 55 subjects, score_7 + score_8 + score_9 + score_10 <= 3:    9744 (0.9744 %)
- 50 subjects, score_7 + score_8 + score_9 + score_10 <= 3:    19216 (1.9216 %)


Edited by stv014 - 5/1/12 at 2:34am
post #21 of 256
Quote:
Originally Posted by kiteki View Post
Example - If I take an original CD, and then create a duplicate CD which went through an ADC process, rip them both to my computer, put the files on a Hifiman HM-601, perform an extensive ABX - and can't tell the difference, what did I just prove?

 

Sometimes your posts are sort of cryptic, but I think we're in agreement here.

All you prove here is that the DAC/ADC chain is more transparent than that specific setup can detect. IE: nothing.

 

 

The thing that bothers me is that he compares his amp and DAC to the benchmark DAC1 and concludes that their both sonically "transparent" and they sound alike.

While no one will say that the benchmark DAC1 sounds bad, I haven't heard many reviews considering it to be "reference".

 

In fact, most people consider the headphone amp in the DAC1 to be quite mediocre. If you tell me you can't tell the difference between it and your own amp.. then I'm slightly confused here.. is that suppose to be a good thing?

 

Now, if a $100 DAC can stand up against the benchmark DAC1, that is saying something.. (something good)
But your design just happens to sound just like another DAC using a completely different chip, design, implementation and budget?

What now? That just doesn't make sense.

 

Two results possible conclusions:

  1. Both your DAC and the other DAC you happen to own are both equally transparent, and transparent in exactly the same way.
  2. Your setup or ears or bias is limiting your ability to tell the difference.

 

 

I'm sorry, but number (2) seems much more likely.


Edited by nullstring - 5/1/12 at 12:58pm
post #22 of 256
Thread Starter 
Quote:
Originally Posted by nullstring View Post

But your design just happens to sound just like another DAC using a completely different chip, design, implementation and budget?

What now? That just doesn't make sense.

 

Why would that matter?

 

The only thing that matters is the output and they're both designed to do the same thing.  If they're both perfect to below the threshold of human hearing than why should they sound different?

post #23 of 256
Quote:
Originally Posted by maverickronin View Post

 

Why would that matter?

 

The only thing that matters is the output and they're both designed to do the same thing.  If they're both perfect to below the threshold of human hearing than why should they sound different?

 

I would agree. If they're both perfect.

What I am saying is that is a very bold statement... that both the benchmark DAC1 and his DAC are perfect...

post #24 of 256

I think its generally accepted that the Benchmark is a very good DAC with a reasonable headphone amp - the old reviews of the Grace M902 seemed to tip the balance in the other direction. I'm more than a little 'meh' on such conclusions - consider this from John Darko's unpublished review of the Centrance Audiophile Desktop:

 

 

The AKG K-702 are a notoriously fickle customer.  Get the amplification wrong and they sound thin and un-involving.  The DACMiniPX goes where other all-in-one-rs (Anedio D1, Peachtree iNova/iDecco) have previously come up short.  The dynamically thrilling joyride of Squarepusher's new release sees Tom Jenskinson is firmly back in the electronica camp.  If you were disappointed by Daft Punk's Tron Legacy soundtrack, Ufabulum is intense, playful and spasticated!  Letting it run on endless repeat it was downright obvious that the AKG don't go undernourished with the PX.  The CEntrance feedbag easily matches the solidly-made and solid-sounding Burson HA-160 for drive and musical saturation.  Using the right-hand socket on the Australian, listener satisfaction is achieved on both amps with the volume placed squarely at midday.  They are more similar than different: a shorter signal path with the PX direct (no interconnects, y'see) means detail is bright eyed and bushy tailed.  Instead, sound routed via Burson + WLM interconnects surrenders a smidgen of minutiae extraction in favour of some transient rounding and warmth.  Swapping CEntrance for Schiit on the DAC front brings more fullness/fatness to the yard.  Head-fi-ers who like 'em big n bouncy will dig the Schiit/Burson combo the most.  Listeners who crave the scientific blade of neutrality will prefer the PX as an all-in-one.  Somewhere in the middle lies the PX + HA-160 combination.
 
That the D1 doesnt do a good job driving the AKGs is news to me, but that looks very much like a plug for the separate DAC/amp combination over the integrated Centrance DacMini PX - in any case, it would involve buying a lot of kit to be able to confirm Darko's findings. What does this have to do with the ODAC ? To my way of thinking, it needs to be assessed on its own merits, not in combination with the O2 - completely disregard every claim made for the DAC and put it with as many amps as possible - speaker amps as well as headphone amps. Be ruthless and be honest - whatever the measurements, if it doesnt sound good with your music/ears/gear, dont be afraid to say so. I'd be keen to hear it alongside my MSII, the DACPort, Bifrost and a host of other entry-level DACs - interesting times ahead. 
 
(fwiw, I wouldnt put too much stock in Darko's conclusions above - he is now on the freelance staff list at 6Moons and that seems like a natural fit for mine)
post #25 of 256
Quote:
Originally Posted by nullstring View Post

 

I would agree. If they're both perfect.

What I am saying is that is a very bold statement... that both the benchmark DAC1 and his DAC are perfect...

 

Agree 100 per cent, but his argument was that the Benchmark had excellent measured performance - backed up by Stereophile's own testing. If there is one point that Voldemort makes which resonates with me, its this : if I want to make any marketing claims about my new SuperDuper HP-10000, and link to glowing reviews from the 'right' people, surely I can provide meaningful specifications alongside that marketing ? I may not have a dScope, but publishing nothing beyond a few power figures (often meaningless in isolation) and glossy images doesn't help anyone. I would link a glaring example of that sort of product, but given that its from a Head-Fi sponsor I wont go there. 

post #26 of 256
Thread Starter 
Quote:
Originally Posted by nullstring View Post

I would agree. If they're both perfect.

What I am saying is that is a very bold statement... that both the benchmark DAC1 and his DAC are perfect...

 

I never said anything was absolutely perfect.  I said if they're both "perfect" to a level below what humans can perceive then they should sound the same.

 

The concept makes sense right?  If the levels of any nonlinearity, noise, distortion, etc of two devices are below what is audible then what we hear must sound the same.

 

Regardless of what this level may be would you agree that such a thing exists?

post #27 of 256
Quote:
Originally Posted by maverickronin View Post

 

I never said anything was absolutely perfect.  I said if they're both "perfect" to a level below what humans can perceive then they should sound the same.

 

The concept makes sense right?  If the levels of any nonlinearity, noise, distortion, etc of two devices are below what is audible then what we hear must sound the same.

 

Regardless of what this level may be would you agree that such a thing exists?

 

 

That's what I meant. (It's annoying to write "perfect to a level below what humans can perceive" in every time I write "perfect").

 

I don't know if I'll agree "such a thing exists".

But I will say that if two devices sound that same within human tolerances, then they sounds the same.

 

But do we know what that is? Does everyone hear sound exactly the same way?

Does every human have the same tolerances for every single thing?

Can we measure every single thing?

Are measurements the only things that matter 100% of the time?

 

I'd just like to point out that I am picking at the difference between the following two..

 

Measures very well -> sounds very good -> sounds about as good as another amp that measures about as well.

I might be able to buy that.

 

Measures very well -> sounds very good -> sounds indistinguishable with another amp that measures about as well or better as long as it's above so called "human tolerances"

This is a much bigger step. This means that measured performance IS audible performance and characteristics in all cases.

 

 

So, to reiterate

I would agree. If they're both perfect to a level below what humans can perceive

What I am saying is that is a very bold statement... that both the benchmark DAC1 and his DAC are perfect to a level below what humans can perceive...


Edited by nullstring - 5/1/12 at 2:14pm
post #28 of 256
Quote:

Originally Posted by nullstring View Post

 

But do we know what that is? Does everyone hear sound exactly the same way?

Does every human have the same tolerances for every single thing?

Can we measure every single thing?

Are measurements the only things that matter 100% of the time?

 

There is an obvious answer to this, but the thread would likely be moved to "Sound Science" tongue_smile.gif

post #29 of 256
Originally Posted by nullstring View Post
Originally Posted by kiteki View Post

Example - If I take an original CD, and then create a duplicate CD which went through an ADC process, rip them both to my computer, put the files on a Hifiman HM-601, perform an extensive ABX - and can't tell the difference, what did I just prove?

 

Sometimes your posts are sort of cryptic, but I think we're in agreement here.

All you prove here is that the DAC/ADC chain is more transparent than that specific setup can detect. IE: nothing.

 

Yes, water is only as transparent as the ink it has to travel through.

 

Audio is like looking through seven windows.  It's intuitive that we're not sophisticated enough to have made seven clear windows yet.  The ODAC and O2 articles have posited that the fifth and sixth windows are 100% transparent, and only the last window is [always] coloured.

 

If you assume or theorise that the fifth, sixth and seventh windows are totally transparent in the study, then you can assume or theorise that the fourth window (the ADC process) was transparent according to the statistics revealed in this specific ABX study.

 

A professor in applied statistics read the entire study, and arrived at this conclusion - 1. "There were approximately 50 subjects and only three managed to score 7/10 or higher? That is indeed a thin tail. Too thin.", and 2. "by my calculation it seems possible that a chi-squared test would reveal that the recording engineers tested significantly better than other respondents. This would be evidence that "golden ears" can distinguish the two formats."

 

Note, I haven't covered points such as:

- The answers solicited from the test conductor / author are unprofessional.

- The controversy and conflicting data of ABX methodology.

- The test aimed at comparing PCM to DSD, and failed to provide DSD content in all testing sessions.

post #30 of 256
Quote:
Originally Posted by stv014 View Post

 

There is an obvious answer to this, but the thread would likely be moved to "Sound Science" tongue_smile.gif

 

Thats going to happen anyway - you know it, I know it and the mods know it. 3 near-identical threads on the ODAC in the same timeframe, and they will all end up in the same place. I just smile and wave.  

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › The Hopelessly Derailed ODAC/Objective DAC Anticipation/Discussion Thread