Lipshitz has said as much also. Here's the quote:
"The amount of negative feedback used in a 1-bit sigma-delta modulator striving to straighten its quantizer transfer characteristic, and simultaneously achieve a signal-to-noise ratio of 120 dB, far exceeds anything ever used before in high quality audio design. Ironically, while a part of the industry mistakenly espouses low feedback for top quality, what we have here is the exact opposite touted as being even better!"
That quote is from this paper.
"The amount of negative feedback used in a 1-bit sigma-delta modulator striving to straighten its quantizer transfer characteristic, and simultaneously achieve a signal-to-noise ratio of 120 dB, far exceeds anything ever used before in high quality audio design. Ironically, while a part of the industry mistakenly espouses low feedback for top quality, what we have here is the exact opposite touted as being even better!"
That quote is from this paper.
I'm not an SACD fanboy. But I did not know how noise shaping modulators worked.
I just find it hard to believe that it has been scientifically proven that SACD and CD can't be told apart.
By implication this means that if Lipshitz's scientific paper proves that if a "1-Bit Sigma-Delta Conversion is Unsuitable for High-Quality Applications" neither is CD suitable for High-Quality Applications....because there is no difference between the sound of CD and that of SACD.
Hence .....Vinyl RULES.
When SACD and DVD-A were first introduced a number of papers quickly appeared proving beyond any doubt that one format was clearly superior to the other.
Otherwise smart people getting distracted by how many angels could fit on the head of a pin.
The first DVD-A players gave only token acknowledgment to the analog output stage. The designers had (mistakenly) assumed that all that was needed was more bits. The earliest SACD releases instead came with a first class player, the Sony SCD-1 quickly followed by an Accuphase player. Both these players were quite careful in their analog sections causing many audiophile (subjectivists) to proclaim SACD as the better format.
Today, even $19.00 computer CD mechanisms produce zero bit errors. All the considerable improvement in CD sound over the last 20 years or so was by way of the reconstruction circuits, Clock, DAC, filter and buffer. If you don't get that right no amount of format "improvement" will affect the sound quality.
Otherwise smart people getting distracted by how many angels could fit on the head of a pin.
The first DVD-A players gave only token acknowledgment to the analog output stage. The designers had (mistakenly) assumed that all that was needed was more bits. The earliest SACD releases instead came with a first class player, the Sony SCD-1 quickly followed by an Accuphase player. Both these players were quite careful in their analog sections causing many audiophile (subjectivists) to proclaim SACD as the better format.
Today, even $19.00 computer CD mechanisms produce zero bit errors. All the considerable improvement in CD sound over the last 20 years or so was by way of the reconstruction circuits, Clock, DAC, filter and buffer. If you don't get that right no amount of format "improvement" will affect the sound quality.
Bas Horneman said:
🙂 One thing that's been pointed out about the Lipshitz paper and others like it is that 1-bit sigma-delta is one possible implementation of DSD, but that DSD is a format not tied to any specific implementation. Hawksford came up with a paper related to this called "Parametrically Controlled Noise Shaping in Variable State-Step-Back Pseudo-Trellis SDM". Apparently this was intended to address some of the theoretical shortcomings mentioned by Lipshitz and others, but I don't know if there are any implementations of this actually being used.
Myself, I have some great-sounding SACDs, but to actually determine on a factual basis whether this format is audibly superior is a really hard problem!
Mmm. Since I'm not planning on becoming a member of AES.
Could you tell me what the central messages of the particular paper is?
Basically, none of their listener could hear the effect of a 16/44.1 A/D->D/A into the signal chain. Systems included extra-loud studio stuff, modern "engineered" hifi, and high-end ESL-based. The tests appear to be well-documented and easily subject to replication and (if necessary) correction.
They are very careful to say in their conclusion that they have NOT proved that the insertion of a 16/44.1 link into a system is never audible to anyone no matter what, but they do reasonably (IMO) claim that, given the good size statistical sample (which they helpfully break down) and the wide variety of systems and listeners where the null result was obtained, the burden of proof is now on those who are claiming SACD's sonic superiority.
They are very careful to say in their conclusion that they have NOT proved that the insertion of a 16/44.1 link into a system is never audible to anyone no matter what, but they do reasonably (IMO) claim that, given the good size statistical sample (which they helpfully break down) and the wide variety of systems and listeners where the null result was obtained, the burden of proof is now on those who are claiming SACD's sonic superiority.
Bas Horneman said:
A few random (but significant) quotes...
"...we felt the need to go further and perform a thorough, straightforward double-blind level-matched listening test to determine whether 16/44.1 technology would audibly degrade the sound of the best high-resolution discs we could find."
"...we have analyzed all of the test data by type of music and specific program; type of high-resolution technology; age of recording; and listener age, gender, experience, and hearing bandwidth. None of these variables have shown any correlation with the results, or any difference between the answers and coin-flip results."
"...it is very difficult to use negative results to prove the inaudibility of any given phenomenon or process. There is always the remote possibility that a different system or more finely attuned pair of ears would reveal a difference. But we have gathered enough data, using sufficiently varied and capable systems and listeners, to state that the burden of proof has now shifted. Further claims that careful 16/44.1 encoding audibly degrades high resolution signals must be supported by properly controlled double-blind tests."
I still wonder why a part of the audiophile audience is simply rejecting/dismissing these results. So far, I have not heard a shred of convincing argument against the author's conclusions and as of today no comments were added so far to the special section (dedicated to this topic) on the AES website.
Bas Horneman said:
Because it wasn't. If it were better, the insertion of the CD standard would have degraded the SACD. It's a nice experimental design.
soongsc said:
Yes, in a follow up on the Boston Audio Society website:
http://www.bostonaudiosociety.org/explanation.htm
From the same website, an interesting (nothing entirely new, though) lecture about ABX testing, by E. Brad Meyer:
http://www.bostonaudiosociety.org/bas_speaker/abx_testing.htm
I could only find measusrments for the Quad ESL 989. And I can say these are a poor choice for this kind of testing. A more stringent criteria should have been used during selection of equipment.
1. Speaker CSD should decay at least 12db within 0.4ms for frequencies above 2KHz.
2. Speaker impedance should not vary for more than 20% for frequencies above 2KHz, and preferrably such even down to 100Hz.
If the speaker can meet at least those criteria, then I think the test would be more believable.
1. Speaker CSD should decay at least 12db within 0.4ms for frequencies above 2KHz.
2. Speaker impedance should not vary for more than 20% for frequencies above 2KHz, and preferrably such even down to 100Hz.
If the speaker can meet at least those criteria, then I think the test would be more believable.
From the lower link:
This blind spot is still bizarre to me. In his lecture explaining the ABX protocol Meyer provides an example of how incredibly sensitive test subjects are to even the most subconscious tester bias. The IQ test administrators did not openly express opinion on the expected result, privately held expectation was enough to invalidate them. Yet far from private, the views on the audibility of audiophile minutia held by Meyer and other demonstration/testers is well know to his subjects. He seems completely unaware his example provides a damning refutation to his demonstration.
This blind spot is still bizarre to me. In his lecture explaining the ABX protocol Meyer provides an example of how incredibly sensitive test subjects are to even the most subconscious tester bias. The IQ test administrators did not openly express opinion on the expected result, privately held expectation was enough to invalidate them. Yet far from private, the views on the audibility of audiophile minutia held by Meyer and other demonstration/testers is well know to his subjects. He seems completely unaware his example provides a damning refutation to his demonstration.
Shouldn't puzzle you at all. That's the principle argument for double blind as opposed to blind.
Soongc, if the difference between SACD and CD is so miniscule that it can't be distinguished on Quad electrostats, it's got to be pretty damn small.
Soongc, if the difference between SACD and CD is so miniscule that it can't be distinguished on Quad electrostats, it's got to be pretty damn small.
SY, The CSD difference can sound like a 3~4 db difference. I think this influence is very significant, and significantly masks the results. Whether 3db difference is of any significance to others, I cannot say.
I took a different lesson from it. While a principle argument for DBT, it's not a blanket panacea for tester bias. The IQ testers presumably weren't looking at the subject's answers and prompting them to a specific response. No reason was given to suggest the testers even knew the correct answers and could guide them wrong by intent. Subjects gave a higher incidence of incorrect responses because something of the tester's demeanor at an imperceptible level affected their performance adversely.
Meyer expected a null result for the digital test, his subjects were reasonably communicated that expectation, a null result was delivered. It's a reasonable criticism a conscientious protocol would strive to avoid, which is why these 'demos' always come across to me as punditry.
Meyer expected a null result for the digital test, his subjects were reasonably communicated that expectation, a null result was delivered. It's a reasonable criticism a conscientious protocol would strive to avoid, which is why these 'demos' always come across to me as punditry.
rdf said:
I may not be reading you correctly, but on DBT it seems to me that those who criticise the testing methodology do so on the grounds that the results don't tally with their subjective experience. By all means study the protocol to find any problems, but just saying the tests are flawed because my subjective experience tells me otherwise is not being scientific, it is being subjective.
fredex said:
Nowhere have I questioned DBT in principle. I have raised specific objections or concerns about the design of specific tests. I haven't said a thing about my subjective impressions. You are reading me, or more correctly reading into me, incorrectly.
- Status
- Not open for further replies.