[kiteki]

*sigh*
 
I wish I had time to type up a proper response...
 
I'll just say that I wasn't questioning if the codec study happened or not.  I'm asking if Robert Harley's anecdote about a conveniently dead expert picking up on something that all tests missed has any other corroboration. [/]

 
Ok, keep sighing, the avguide article I linked was from 2008, this forum post is from 1999 - http://www.audioasylum.com/audio/general/messages/57/10967.html
 
"First, Locanthi related his experience of listening to a DAT tape that contained examples of low-bit-rate encoded music. He had requested the tape from Swedish Radio, the organization which conducted the official listening test of these systems. Almost immediately Locanthi heard several peculiar sounds in the music, the most obvious being an idle tone at 1.5khz. When Locanthi imformed Swedish Radio of this problem, they were surprised that they had not discovered it, but they did hear the 1.5khz artifact after it was pointed out to them. When Locanthi asked how such an obvious flaw could go undetected, the response was that he 'knew what to listen for'".
 
Swedish Radio conducted "over 20,000 separate trials and 60 'expert' listeners". They failed to detect a flaw immediately apparent to a single listener. Their listening-test methodology--called 'hidden reference, double-blind, triple stimulus'--was beyond scientific reproach. Yet a single listener in 'unscientific' listening conditions immediately identified this fundamental problem."
 
 

 
[/]
Even if it is true it still doesn't mean much because the test wasn't an ABX test. It was ranking blinded samples. I'd imagine that in 1991 the best lossy compression still sucked.

 
Interesting, let's see what Wikipedia says?
 
"Subjective audio testing by experts, in the most critical conditions ever implemented, has shown MP2 to offer transparent audio compression at 256 kbit/s for 16-bit 44.1 kHz CD audio using the earliest reference implementation (more recent encoders should presumably perform even better).^{[size=x-small][54][/size]"}
 
Did you see the word "transparent"?
 
Do you see a pattern here?  Wikipedia article -> AESJ -> proven statistical error / failed test --- Let me guess, it's psi-missing, ね？
 
If you want to continue the discussion please post the relevant link on my wall.

 

[stv014]

Quote:

Originally Posted by kiteki /img/forum/go_quote.gif
 
"First, Locanthi related his experience of listening to a DAT tape that contained examples of low-bit-rate encoded music. He had requested the tape from Swedish Radio, the organization which conducted the official listening test of these systems. Almost immediately Locanthi heard several peculiar sounds in the music, the most obvious being an idle tone at 1.5khz. When Locanthi imformed Swedish Radio of this problem, they were surprised that they had not discovered it, but they did hear the 1.5khz artifact after it was pointed out to them. When Locanthi asked how such an obvious flaw could go undetected, the response was that he 'knew what to listen for'".
 
Swedish Radio conducted "over 20,000 separate trials and 60 'expert' listeners". They failed to detect a flaw immediately apparent to a single listener. Their listening-test methodology--called 'hidden reference, double-blind, triple stimulus'--was beyond scientific reproach. Yet a single listener in 'unscientific' listening conditions immediately identified this fundamental problem."

 
Some obvious questions, in addition to how the Swedish Radio tests were conducted (they might have been "beyond scientific reproach" regarding false positives, but were the subjects given every possible advantage that a good ABX test offers ?): would Locanthi have found the flaws also in a DBT ? Would the "60 expert listeners" have found it (i.e. specifically the 1.5 kHz artifact, and not dozens of imagined ones) in a test that is sighted but otherwise identical to the Swedish Radio one ? Would they have heard it in a DBT after already knowing what to listen for ? Your example does not necessarily prove that sighted tests have any inherent advantage compared to blind ones, only that Locanthi was a more expert listener than the others, or the DBT was overly restrictive.
 
Edit: the purpose of the "blind" part of the test is not to discover the differences, but to prove that you really hear them. Therefore, the test should allow the subject to listen to both A and B sighted (knowing what they are) with no restrictions, with the only hidden information being the identity of X.
 
Quote:

"Subjective audio testing by experts, in the most critical conditions ever implemented, has shown MP2 to offer transparent audio compression at 256 kbit/s for 16-bit 44.1 kHz CD audio using the earliest reference implementation (more recent encoders should presumably perform even better).^{[size=x-small][54][/size]"}

 
There is an important difference between a codec that can be transparent, and one that is guaranteed to be transparent with any source material. I have tested the original layer 3 "reference encoder" source code in the 90's, and it sometimes had easily audible artifacts even at 320 kpbs. That is why the LAME project was started to improve it and fix its numerous flaws.

 

[kiteki]

 
Edit: the purpose of the "blind" part of the test is not to discover the differences, but to prove that you really hear them. Therefore, the test should allow the subject to listen to both A and B sighted (knowing what they are) with no restrictions, with the only hidden information being the identity of X.

 
"Swedish Radio developed an elaborate listening methodology called “double-blind, triple-stimulus, hidden-reference.” A “subject” (listener) would hear three “objects” (musical presentations); presentation A was always the unprocessed signal, with the listener required to identify if presentation B or C had been processed through the codec."
 
http://www.aes.org/e-lib/browse.cfm?elib=5396
 

---

As for the rest of your post, yes, all those questions look valid.
 
There is an important difference between theoretical transparency and true transparency, yes, I was just citing the incorrect Wikipedia entry ...

 

[estreeter]

Quote:

I think it has a lot of sterility and high performance.  It sounds like a clean hospital... there.

 
Is it the job of the DAC to sound like 'Cheers' bar at 2am on Sam's Stag Night  ?  Surely the speakers/phones should assume that role.  

 

[Tronz]

Quote:

How do you know that? From the DAC IC that its using? On that argument, I know that the ODAC won't be audibly transparent coz I've heard that chip.

It's not about the DAC chip used, it's about it's implementation. According to the ODAC's specs, it's audibly transparent.

 

[khaos974]

What is all this talk about thread closing, wouldn't it be more productive to warn or ban temporarily the members who violated the TOS and highlight the speific section of the TOS they violated so that everyone know no to do so?

As far as I'm concerned I've already heard it but I'll buy it anyway just to make sure they sound the same as an experiment - (with my perception, equipment, statistical error a.k.a. psi-missing, visual-audio illusions a.k.a. psychoacoustics, transparency of my headphone / speaker / IEM, bla ha bla ha ~)

I think it has a lot of sterility and high performance.  It sounds like a clean hospital... there.

The ackodac is not even remotely close the ODAC, one uses a $65 chip, the ES9018, the other uses a $1.5 chip, the ES9023.

 

[Tronz]

Quote:

As far as I'm concerned I've already heard it but I'll buy it anyway just to make sure they sound the same as an experiment - (with my perception, equipment, statistical error a.k.a. psi-missing, visual-audio illusions a.k.a. psychoacoustics, transparency of my headphone / speaker / IEM, bla ha bla ha ~)
 
I think it has a lot of sterility and high performance.  It sounds like a clean hospital... there.

Have you heard the ODAC itself or a board that happened to have the DAC the ODAC uses implemented onto it? If it's the latter, like I've said before, the implementation of the DAC is more important than the DAC chip used. Also, I find it amusing how you guys always associate an accurate sounding equipment with it being sterile, especially if it's at a low price. There could be another DAC that's relatively accurate that costs a lot more and through sighted testing, 95% of the time, the more expensive DAC wouldn't be classified as "sterile".

 

[HaVoC-28]

Quote:

 
Example - "To-99 op-amps sound better", afaik there is no measurement or study to support this, but there are theories of less EMI, RFI or radiated RFI, whatever, and Texas Instruments is making them, why?  In fact can I ask, why is TI making higher performance audio IC's at all, if something like NE5532 is 100% sonically transparent?  Why don't you write a letter to TI and ask!  I've seen at least one response from National.
 

 
For TO99 is easy to understand , TO99 => made to resist to the toughest conditions , DIP8 => easely swappable , standard ?  , Soic 8 => minimum space needed ? , opamp are made for a wide variety of usage ... 
 
And i am not sure asking TI or National directly is the way to go , they are not exactly philantropic ... They will help you a bit with recommanded opamps like for PCM1792A in a given circuitery :
 
"The operational amplifier recommended for the IV circuit is the NE5534, and the operational amplifier
recommended for the differential circuit is the Linear Technology LT1028, because its input noise is low."
 
http://www.ti.com/lit/ds/sles105b/sles105b.pdf
 
They will argue that a dermided opamp is better suited to some circuitery etc etc . The best from an electrical stand point , but for sound ... 

 

[Currawong]

Quote:

Quote:

Tronz: Switch "documentation" for "marketing". 

Didn't know someone would make up specs as a marketing tactic when they receive zero money from the project. 

 
I never said anything was made up. You made that up. 

 
Marketing is about making something appealing so people will accept and buy it.  Think about why the O2 and ODAC are appealing to some. That's all I'm saying. 

 

[estreeter]

Quote:

What is all this talk about thread closing, wouldn't it be more productive to warn or ban temporarily the members who violated the TOS and highlight the speific section of the TOS they violated so that everyone know no to do so?
.

 
Not closed - just moved to Sound Science, where it will die a natural death. Its already dead to me - Sound Science is a great place for metaphysical discussions about something no-one has actually heard. 

 

[khaos974]

What is all this talk about thread closing, wouldn't it be more productive to warn or ban temporarily the members who violated the TOS and highlight the speific section of the TOS they violated so that everyone know no to do so?

.

Not closed - just moved to Sound Science, where it will die a natural death. Its already dead to me - Sound Science is a great place for metaphysical discussions about something no-one has actually heard. 

It deserved to be moved, the discussion shifted to peer reviewed papers and limits of audibility, but there's still kiteki's thread; http://www.head-fi.org/t/607975/epiphany-acoustics-e-dac-jdslabs-odac-head-n-hifi-odac-discussion

 

[estreeter]

Just give me the weekend to take care of that straggler ......   

 

[maverickronin]

Quote:

Ok, keep sighing, the avguide article I linked was from 2008, this forum post is from 1999 - http://www.audioasylum.com/audio/general/messages/57/10967.html

 
And it goes back to Harley telling the same story in Stereophile...
 
Quote:

 
"Swedish Radio developed an elaborate listening methodology called “double-blind, triple-stimulus, hidden-reference.” A “subject” (listener) would hear three “objects” (musical presentations); presentation A was always the unprocessed signal, with the listener required to identify if presentation B or C had been processed through the codec."
 
http://www.aes.org/e-lib/browse.cfm?elib=5396

 
I can't afford to shell out for every paper someone mentions but the first page mentions using a grading scale.  That's not the kind of methodology you use to to detect absolute differences.

 

[mikeaj]

I can't afford to shell out for every paper someone mentions but the first page mentions using a grading scale.  That's not the kind of methodology you use to to detect absolute differences.

The IEEE Xplore link is only to two pages, the second needing access. It looks like the IEEE version is for a different conference than the AES submission, though the title and authors are the same. The content is probably different as well.

The rest of the "Assessment Procedure" section is (since I have access):

It was also agreed to use the Listening Test System, designed and built by SR in 1985. The system minimizes the influence of the human error factor and guarantees an unbiased result.

The basis for the selection of test material was to include 2 critical test sequences for each codec, plus 2 sequences for evaluating possible stereo image shifting. The sequences should be "normal broadcast material" and not artificial signals.

It was agreed to use approximately 50 expert listeners in the tests.

Then it goes on to summarize results for ASPEC compared to CD quality, and MUSICAM compared to CD quality, and what MPEG decided on.


Almost all research on subjective impressions in audio are for situations where there are actually differences to be heard, like for low bitrate encoding or maybe bit errors during real-time transmission of audio (which requires some interpolation to cover for the missing data). These days there are some standards for grading scales and so on. There's much more interest in finding "good enough" rather than "transparent" as one would expect.

 

[AKG240mkII]

Quote:

I don't know whether to laugh at this thread or ban you all. I'll decide tomorrow morning (it's 11pm here). 

 
Some things to consider: Measurements are almost all made using sweeps, tones and impulses. This is not the same as playing music.  Also, while it's easy for a lot of gear to sound good with a lot of music, try playing very complex and dynamic orchestral music (I use Eiji Oue's Bolero! Orchestral Fireworks) and see if the sound doesn't compress or blur. Cheap gear usually fails this for much the same reasons in a source as an amp -- both contain amplification stages of some kind and if they are insufficient, can't keep up with the continual rapid changes in voltage. The DX100, since you guys have been mentioning it, sounds a bit blurry playing complex classical. I know though that if I were listening to a simple jazz piece using the DX100 as a source and comparing it to my main DAC connected to my Stax rig, it is much harder to tell them apart. The difference in detail compared to an Esoteric K-01 is much more apparent, however.
 
Oh, by the way, that Linn amp ruined me. I want it more than a Blue Hawaii. It is that good. There are almost none available second-hand anywhere.

Right, your uber-amp is somehow aware of the fact that it's a cello, not a snare-drum, and because of some magic 'synergy' it amplifies them differently ?
There is no such thing as 'music' until your speakers start to make air-molecules vibrate .
 
My background is 30 years plus in the business that produces the stuff you listen to : Music !
We have nothing but contempt for all your insane, unfounded HiFi voodoo-******** .
There is a reason you don't see 2000 bucks pr meter 'cables' in recording-studios !
 
But let's not discuss it .. Let's ABX it !!!!