There's a fourth possibility as well, which you were too polite to bring up. But I'm working off the hypothesis that #1 is the answer and I'm trying to get enough from Charles to understand what the rational explanation is. It's still unclear what the "pairs" were, but if they are digital files (whether data or software), it costs him nothing to send them to me, and perhaps something useful and interesting will result.
... then again, if there were arguments instead of the swear words, we'd all be convinced mightily by now
jan didden
Are you kidding? Have you bothered to actually read anything I've written about this?
42.
Digital File Errors
Agreed on this. There is a huge difference between different software's ability to adequately play HDD files. In some cases (as I have seen with Winamp) merely increasing the program's buffer fixes the issue. In other cases, the quality differences are due to some program feature (bug) that cannot be circumvented (WMP-just try to get it to play VBR mp3!). Sometimes it is due to underlying data transfer issues due to CPU utilization, bus congestion or HDD sub-system latency.
It is possible that two files that pass a bit-for-bit equivalency test on the same HDD can play differently if one is fragmented on the HDD and causes buffer depletion.
A few years ago before SATA performance became adequate we had to use Ultra320 SCSI in our mastering machines to keep transfer rates high enough to eliminate data corruption in our glass masters at 8x mastering speed, even though no errors were reported in the process, they were found by bit-for-bit comparisons of the resulting glass with the source. This proves something I have always hated about digital audio, especially that done in the PC or Mac domain: error detection, correction and reporting is entirely determined by the will and skill of the coder, not by any nature of the medium. This means belief in digital's perfection on playback is largely matter of faith, not fact. The firmware in CD mastering systems and players has largely been improved and proofed over the last 30 years. Still....
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
www.wxyc.org
1st on the Internet
Agreed on this. There is a huge difference between different software's ability to adequately play HDD files. In some cases (as I have seen with Winamp) merely increasing the program's buffer fixes the issue. In other cases, the quality differences are due to some program feature (bug) that cannot be circumvented (WMP-just try to get it to play VBR mp3!). Sometimes it is due to underlying data transfer issues due to CPU utilization, bus congestion or HDD sub-system latency.
It is possible that two files that pass a bit-for-bit equivalency test on the same HDD can play differently if one is fragmented on the HDD and causes buffer depletion.
A few years ago before SATA performance became adequate we had to use Ultra320 SCSI in our mastering machines to keep transfer rates high enough to eliminate data corruption in our glass masters at 8x mastering speed, even though no errors were reported in the process, they were found by bit-for-bit comparisons of the resulting glass with the source. This proves something I have always hated about digital audio, especially that done in the PC or Mac domain: error detection, correction and reporting is entirely determined by the will and skill of the coder, not by any nature of the medium. This means belief in digital's perfection on playback is largely matter of faith, not fact. The firmware in CD mastering systems and players has largely been improved and proofed over the last 30 years. Still....
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
www.wxyc.org
1st on the Internet
As Rafiki told the young Lion King, "L O O K C L O S E R R R..."
Those are not op-amps. An op-amp (by definition) operates by using feedback. Our 7-series products use Analog Devices AD844's (designed by the great, but very stubborn, Barrie Gilbert) in the analog signal path, but with no feedback. So while the part was designed to be used as a current-feedback op-amp, we don't use it that way. Just one reason that they are in the "Class A" rating in Stereophile.
You own it, so you tell me if it sounds "high-end" or not.
Barrie would be with the know-nothings on this. If #1 is the answer why couch the problem as essentially violating first principles? Sorry that Heisenberg, Shannon, et al are wrong, in the case of audio, will not figure into your answer.
It has been said elsewhere that somehow data files "remember" their history and it is contained in them in some unknown mysterious way that is not yet known. Yawn.
Good luck SY, probably transmitting the files will break the mojo.
Last edited:
While i understand all the concerns, the question is if someone takes the hardship of a 3 trial test, why not just add 2 or three more trials to get past the normally accepted level of significance barrier of 0.05?
In the one tailed case a number of 5 trials brakes that barrier (actual p=0.031). In the two tailed case you´d need 6 trials to get down to p=0.031 .
Of course a controlled listening test isn´t really fun, and it is not like flipping a coin, but unfortunately the statistics is the same as for flipping a coin.
If you are testing pairs and choosing one or the other three times in a row is a valid result, then the chance for guessing is 1:4 but not 1:8 .
Scott & SY,
This is one of those items where we clearly disagree. My read on this is we have an anecdotal story of being able to distinguish different so called perfect signals WITHOUT knowing all of the exact details. So is the deed true? My OPINION would be yes. What was the difference? Don't know, that information is best known by the programmer. We can't even agree on what an error less file is. Is it the actual file or the error corrected final product?
Now is it reasonable that a programmer insisted his work was proper and others did not agree? Then when motivated by some demonstration of perhaps magical powers went back to work and changed things? A reasonable explanation but again we do not have sufficient information to determine very much.
ES
Short memory. Better cut out the aluminum in your cookware. In brief, DBTs are routinely used in wine evaluation. I spent 12 years of my career designing and performing these tests. To qualify for any sort of certification (MW, MS, WCET) or to qualify as a judge at competitions or for sensory panels, candidates must take and pass a series of DBTs.
So, your analogy (for the 50th time, at least) is thoroughly incorrect and based on a totally false premise, which you've been told again and again is false. 42.
No, he showed how little was generated by a made up 40yr old style op-amp. As I said before "serious" never entered into it, he finds people quoting his article this way quite silly.
To jog your memory the real tempest was over the in-harmonic frequencies, another impossibility.
Last edited:
Well, I don't know, either- if the same bits are delivered at the same time to the DAC, then claims of audible differences are suspect. If the claims of audible differences are correct, then there's a reason and that's why I've asked (several times) for the files in question.
- Status
- Not open for further replies.