CD Errors
Well said. (insert chest thump here) As Director of Engineering at a large optical disc replication plant since 1993 I have gone through explaining this to hundreds of customers, some who really should know better. If an industrial bit-for-bit comparitor such as Eclipse ImageVerify ($15,000/seat) shows a bit equivalence between two discs or a disc and a HDD file when run twice (source and target swapped between verifications) then the two are digitally identical, and in my 18 years of CD molding and listening experience will always sound identical.
There are indeed large differences in sound quality between players which are not interpolating due to circuitry differences, that is not the issue. I have heard large differences between two discs from the same replication run...on two different players, NEVER on the same player. It is then easy enough (once you get the decoder chip schematic) to find the C3 or concealment flag pin and verify that there are indeed uncorrectable errors on one or both players resulting in interpolation. When this happens on a disc that passes all Sony/Philips Red Book parameter checks, BY VERY DEFINITION it is a player error, not a disc error. This is because the definition of a correctly designed player is one that will correctly play ANY Red Book compliant CD. The entire argument about CD bits and sound quality can be reduced to a short paragraph:
Any difference between a source and replica disc once the replica has passed both a bit-for-bit and RedBook parameter check are by definition due to player issues causing actual uncorrectable errors which are then interpolated, resulting in the sound difference.
I have had plenty of audiophile customers tell me that they can hear the difference between two HDD files which are bit-for-bit equivalent at which point I ask: "Awrighty then; If you listen to the same file twice is there a difference as well?" because that would be the same test if indeed there was a bit equivalence. Amazingly some have tried that and told me that there is a difference, to which they conclude their systems are unstable. I believe that sensory, mental, and environmental changes from one listening to another are so gross compared to any other possible explanation that they swamp the equation...Occams' Razor.
I keep this in the back of my mind whenever I hear discussions regarding AB comparisons as well...
Reety awrighty I'm the Zombie Woof...and just my $0.02 worth.
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
www.wxyc.org
1st on the Internet
Well said. (insert chest thump here) As Director of Engineering at a large optical disc replication plant since 1993 I have gone through explaining this to hundreds of customers, some who really should know better. If an industrial bit-for-bit comparitor such as Eclipse ImageVerify ($15,000/seat) shows a bit equivalence between two discs or a disc and a HDD file when run twice (source and target swapped between verifications) then the two are digitally identical, and in my 18 years of CD molding and listening experience will always sound identical.
There are indeed large differences in sound quality between players which are not interpolating due to circuitry differences, that is not the issue. I have heard large differences between two discs from the same replication run...on two different players, NEVER on the same player. It is then easy enough (once you get the decoder chip schematic) to find the C3 or concealment flag pin and verify that there are indeed uncorrectable errors on one or both players resulting in interpolation. When this happens on a disc that passes all Sony/Philips Red Book parameter checks, BY VERY DEFINITION it is a player error, not a disc error. This is because the definition of a correctly designed player is one that will correctly play ANY Red Book compliant CD. The entire argument about CD bits and sound quality can be reduced to a short paragraph:
Any difference between a source and replica disc once the replica has passed both a bit-for-bit and RedBook parameter check are by definition due to player issues causing actual uncorrectable errors which are then interpolated, resulting in the sound difference.
I have had plenty of audiophile customers tell me that they can hear the difference between two HDD files which are bit-for-bit equivalent at which point I ask: "Awrighty then; If you listen to the same file twice is there a difference as well?" because that would be the same test if indeed there was a bit equivalence. Amazingly some have tried that and told me that there is a difference, to which they conclude their systems are unstable. I believe that sensory, mental, and environmental changes from one listening to another are so gross compared to any other possible explanation that they swamp the equation...Occams' Razor.
I keep this in the back of my mind whenever I hear discussions regarding AB comparisons as well...
Reety awrighty I'm the Zombie Woof...and just my $0.02 worth.
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
www.wxyc.org
1st on the Internet
Last edited:
Ah, yes, but British aristocracy is not about quality at all, really.
More to do with a statement of sorts, old chum.
Tada, cheerio, and all that.
The point is that Charles got the sequence right, as I got the line stages right, decades before. This is considered 'coincidence' by many, but for us, it is a day's work. All we care about is making the 'right' decisions to keep our products competitive in the marketplace with others who are putting out the same effort.
So the discs contain errors which are corrected as they are played and this should result in error free final files. If two discs produce the same final file then if the CD player is working correctly it should produce to the limits of it's ability the same output.
So in the case of a non-perfect error correcting CD player two final processed perfect files can sound different.
Therefore the observations of identical files sounding different are actually possible!
The question now becomes when files are compared are they compared as raw data or as final corrected files?
Scott, did I get you are relying on anecdotal evidence? One of my first DSP only sound systems crashed almost weekly, just at the time predicted by memory bit rate errors. The penalty for failure to perform at the first game was $2,000,000. Reboot time was under a minute.
Last edited:
Listening Tests
There seems to be a lot of interest and a lot of confusion regarding the listening tests I performed with computer audio playback.
I wasn't listening to different files. I was listening to different software programs playing back those files.
There is one group of know-nothings that says they couldn't sound different unless the bits delivered to the DAC were different. Well, the bits were the same. That is why I performed the test -- to convince the programmer that the way his program delivered the bits could make an audible difference.
There is another group of know-nothings that claims that there could be an audible difference, but only because the two programs had different amounts of jitter in the output data stream. Wrong. The DAC in question was an asynchronous USB DAC with the master clock in that box. There is a buffer for the incoming audio data and a feedback pipeline that tells the computer to send data more quickly or less quickly, depending on how full the buffer is. There is no contamination from the computer or ground loops because the DAC has total galvanic isolation from the computer via opto-isolators.
I have no idea why the two programs sounded different. I just know that they did, and I proved it with a blind test.
There is a third group of know-nothings that says "Oh, well, one in eight chances is not statistically significant. The result is meaningless." To them I say "Go bugger off." The result would have been the same with three trials or three hundred trials. But it's not like flipping a coin. It actually takes work to do these tests. I might spend an hour or two listening to be sure of my decision. I have better things to do than waste my time doing more trials. Why? Because no matter how many I do, it would never be enough.
If I did 1000 trials and got them all right, this group would say, "Well, he is just an outlier and a normal person could never hear the difference" or some other namby-pamby excuse. Like I said, go bugger off.
There seems to be a lot of interest and a lot of confusion regarding the listening tests I performed with computer audio playback.
I wasn't listening to different files. I was listening to different software programs playing back those files.
There is one group of know-nothings that says they couldn't sound different unless the bits delivered to the DAC were different. Well, the bits were the same. That is why I performed the test -- to convince the programmer that the way his program delivered the bits could make an audible difference.
There is another group of know-nothings that claims that there could be an audible difference, but only because the two programs had different amounts of jitter in the output data stream. Wrong. The DAC in question was an asynchronous USB DAC with the master clock in that box. There is a buffer for the incoming audio data and a feedback pipeline that tells the computer to send data more quickly or less quickly, depending on how full the buffer is. There is no contamination from the computer or ground loops because the DAC has total galvanic isolation from the computer via opto-isolators.
I have no idea why the two programs sounded different. I just know that they did, and I proved it with a blind test.
There is a third group of know-nothings that says "Oh, well, one in eight chances is not statistically significant. The result is meaningless." To them I say "Go bugger off." The result would have been the same with three trials or three hundred trials. But it's not like flipping a coin. It actually takes work to do these tests. I might spend an hour or two listening to be sure of my decision. I have better things to do than waste my time doing more trials. Why? Because no matter how many I do, it would never be enough.
If I did 1000 trials and got them all right, this group would say, "Well, he is just an outlier and a normal person could never hear the difference" or some other namby-pamby excuse. Like I said, go bugger off.
I agree that different SW sounds different playing the same file. Same experience.
No wonder, look here:
SRC Comparisons
No wonder, look here:
SRC Comparisons
Now your mad, OK. What did the programmer think? You usually have a group of folks around who all hear the same thing. How come it never happens that anyone's around that says, "OK so what, I hear nothing". You must filter out all the mid-fi ears.
Next time you get MRI or CAT scan you can thank all us know-nothing engineers who choose to remain ignorant.
Last edited:
I think that something that I mentioned on another thread might play into this a bit. One of our Audio Club members runs a SOTA Recording/Mastering studio (He processes & assembles all the High-Rez work for Chesky/HDTracks) was playing around and burned two disks off of his files. One disc was burned with a Plextor burner and the other was burned using an LP Blu-Ray Burner that he picked up at Fry's. Both were burned on discs from the same stack. To make this a little shorter, he was amazed that they sounded different and upon checking found that they were bit for bit identical. He took the Discs to the U. of Washington Medical Center (his "other job") and looked at them under an electron microscope. The Plextor disc had pits with irregular edges, that appeared frayed and tattered, while the LP Blu-Ray disc had pits which were clean, sharp and well formed.
He said that the Blu-Ray disc was, to him, far better sounding and while he could see a difference, they always tested as identical.
It may be that the tolerances and precision needed for the higher density of information on Blu-Ray discs accounted for what was, upon examination, a much cleaner burn.
I might add that there have been many people (on different Continents) reporting that various bit-identical recordings have sounded different.
I'm no expert, so I'll leave it to those that wish to speculate upon what's happening here.
FWIW, I could hear a difference in the two discs, and I certainly would never claim to have "Golden Ears."
Best Regards,
TerryO
So what were the "test pairs?" Three pairs of software programs? If so, I'd still like to get your test pairs to understand what the differences are.
Charles got it right. It is for those who hear the difference that the success goes to, not those who cannot.
What if someone came to you and INSISTED that virtually all cheap wines tasted the same or at least of the same quality as more expensive wines almost all the time? Would you believe him, and settle for 2 buck Chuck, evermore? I doubt it. Yet there are many people who are convinced of this, and think that wine differences are a 'joke'. Sound familiar?
I am not a wine connoisseur, but I have tasted fine wines. I could be fooled, but not always. Yet, although I do not seek to spend time and effort in making or drinking fine wines, but I do NOT criticize those who do so. Charles and I are like 'winemakers' and we HAVE to succeed in make something exceptional, or we fail in the marketplace. We have been fairly successful over the decades, so maybe we know something that others might overlook.
What if someone came to you and INSISTED that virtually all cheap wines tasted the same or at least of the same quality as more expensive wines almost all the time? Would you believe him, and settle for 2 buck Chuck, evermore? I doubt it. Yet there are many people who are convinced of this, and think that wine differences are a 'joke'. Sound familiar?
I am not a wine connoisseur, but I have tasted fine wines. I could be fooled, but not always. Yet, although I do not seek to spend time and effort in making or drinking fine wines, but I do NOT criticize those who do so. Charles and I are like 'winemakers' and we HAVE to succeed in make something exceptional, or we fail in the marketplace. We have been fairly successful over the decades, so maybe we know something that others might overlook.
No pleasing everyone.
Personally, i find terms as GMAFB(oldy) and GBO rather amusing, shame this box does not do accents also.
Last edited:
Actually MRI was made practical by Raymond Damadian who at first was laughed at. He started using the basic technique for his research, realized it could be made into a scanner and built the first one. So of course the Nobel went to Paul C Lauterbur and Peter Mansfield for their contributions.
It was nice to hear a polite measured rebuttal from Mr. Hansen.
As a mere "know-nothing" can I say that when something impossible happens, one of the following must be true:
1. It wasn't impossible - there is a rational explanation, although we might not know what it is.
2. It didn't happen - for whatever reason, people were fooled.
3. Something supernatural has happened.
We are being asked to believe that identical data with identical timings can sound reliably different. I regard this as a claim that something impossible has happened. Which of the three options above is the best explanation?
As a mere "know-nothing" can I say that when something impossible happens, one of the following must be true:
1. It wasn't impossible - there is a rational explanation, although we might not know what it is.
2. It didn't happen - for whatever reason, people were fooled.
3. Something supernatural has happened.
We are being asked to believe that identical data with identical timings can sound reliably different. I regard this as a claim that something impossible has happened. Which of the three options above is the best explanation?
John, from now on, whenever you bring up that thoroughly incorrect analogy, I'm not even going to bother to shoot it down for the nth time. I'll just say "42." That will save a lot of redundant typing.
42.
- Status
- Not open for further replies.