First of all, error correction does exactly that, correct the error. After correction, there is no more error, it's (again) bit perfect.
If error correction fails, there's a fallback with error concealment which includes interpolation. In general, that IS audible.
But, if that is the case, you have substandard drives or badly scratched disks.
All this has, of course, nothing to do with Save Mode or Normal Mode.
jan
I never trust my senses more than my tools (controlled tests included). Brain can play some funny tricks, especially when you WANT something to be true.
Last edited:
Sorry John, but what you call 'magically' is called 'perception' by the rest of the world.
It is very regretfull that you selected to ignore that area of knowledge for the last few decades.
jan
Finally found a copy of the ST-140 schematic. I only had a hard copy but it went when I sold mine. Just might model it for experience. It did fail the wife test. One more data point. As the "magic" mod was to replace the red LED with green, I can now see what that does to the bias of the IPS.
Jan, when dealing with the crude error correction of a CD, error correction is a best guess, not a true correction like we do with data formats like old 7 of 9 code or GCR on the old tape drives, let alone the application layer methods. No ecc or crc bytes, no retry, no blocking out bad blocks. Dust comes and goes. It is interpolation as you suggest. There is no true error correction, as there is no redundant data from which to extract the correction. You may be surprised that trellis coding for modems corrects only to a statistical probability. We then have the transmission protocol layer request retry if a block does not check after the best guess by constellation position.
We agree on the root problem, poor drive, poor media, poor source. Killing background apps should only have an effect on really really bad software, of which plenty exists! I believe EAC tries to identify read errors and retry.
I'll still take the level of errors induced in a CD copy to those in a tape copy. I would prefer to get my music in a reliable data format that we could employ bit error correction on.
We agree on the root problem, poor drive, poor media, poor source. Killing background apps should only have an effect on really really bad software, of which plenty exists! I believe EAC tries to identify read errors and retry.
I'll still take the level of errors induced in a CD copy to those in a tape copy. I would prefer to get my music in a reliable data format that we could employ bit error correction on.
How's this for the obvious: If you are only trying to satisfy yourself, then it does not matter if you are fooling yourself or not. Ignore measurements as it is only your perception. Be happy, don't worry. But if you are building things for someone else, you had better rely on both measurement and perception as what make them believe may not be the same as you.
Red Book format is very redundant: has EFM encoding (eight-to-fourteen modulation), CIRC (cross-interleaved Reed–Solomon coding), L2 ECC, subcode channels and so on, which are not typically exposed to the application reading the disc. CIRC adds to every three data bytes one redundant parity byte.
Also, any decent software will retry reading a bad data in "data mode", only in "audio mode" there is no retry.
Also, any decent software will retry reading a bad data in "data mode", only in "audio mode" there is no retry.
tvr, that's not exactly true. There are two types of errors- correctable and uncorrectable. In the latter case, there's interpolation. In the former, the original data is restored exactly. So the interpolations are transient events, not continuous. When I had my old Magnavox CD player, I was able to see how many interpolation errors I got. I was surprised by the results. Maybe one or two errors total in ten discs I tried. So out of 11 hours of music, interpolation was used for about 50 microseconds total.
That's in the category of things to not worry about.
why guess when the subject is well covered in depth on the web
Reed-Solomon Codes and CD Encoding
appears competent - any 2 single bit errors in a 24 byte block can be corrected, the blocks are spread out over frames with another error correcting code
due to the interleaving, levels of ecc, up to 3500 sequential lost bits from scratches of the physical media can be corrected
some CD drives give hardware flags for number of ecc corrections, level of the ecc error and finally when interpolation is used (basically never)
no engineering account I know of investigating the "issue", reports of looking at the ecc flags on drives playing CDs supports the idea that there are widespread "major problems" reading CDs
Reed-Solomon Codes and CD Encoding
appears competent - any 2 single bit errors in a 24 byte block can be corrected, the blocks are spread out over frames with another error correcting code
due to the interleaving, levels of ecc, up to 3500 sequential lost bits from scratches of the physical media can be corrected
some CD drives give hardware flags for number of ecc corrections, level of the ecc error and finally when interpolation is used (basically never)
no engineering account I know of investigating the "issue", reports of looking at the ecc flags on drives playing CDs supports the idea that there are widespread "major problems" reading CDs
Precisely!
The question now is - what if it is NOT ripped by "good software"?
How to explain that when I play on the same CD player the original factory copy, it plays just fine, as well as the player can do, but when I put on a copy, made on my PC, the treble socks it to me?
Never mind whether my PC is up to the job or not, the point is that the same copy flaunting all the treble, when played on my Yamaha CDX 993 player used as a transport, and on my external DAC, there is no treble excess? Because my DAC has no brickwall filters, no oversampling, and uses 8 parallel Philips DACs?
Again, never mind which is right or wrong, the key point here is - copies CAN sound different from the original, even assuming very special circumstances. I, and millions like myself, don't really care how and why, or who's to blame, we simply note the differences and, more often than not, do not like them.
To me, the mere statement that something is perfect in an imperfect world is, to be kind, ridiculous. It's childlishly naive.
There is no perfect amp, no perfect preamp, no perfect loudspeaker, no perfect PC - most here would agree, yet some would argue that digital technology as such is perfect.
Sorry folks, but it isn't. It's better than most others, no argument there, but it's NOT perfect. And just so you don't think I'm some kind of a Luddite, let me mention that during the last 25 years, I purchased 0% of LPs and 100% of CDs. I did, do, and will enjoy them even if they are not perfect. And some of the CDs are digital versions of mono analog tape recordings in God knows which forsaken room Pete Seeger recorded it - and to me, he's still THE Big Daddy of protest songs, even in mono.
He doesn't trust his ears. If he would, he would have no problem doing a controlled test.
jan
Frankly, I don't give a FF about controlled tests by anyone. They mean exactly nothing to me because they are 100% irrelevant for me, in my room, and with my system.
If I want to find out how something sounds, I go out there and borrow a sample which has not been tweaked. I use it for no less than two weeks, and only then will I dare say that I KNOW how it sounds (to me). I have only one item in my possession which I did not audition per se, and that was gift, and I daresay a very generous one. It turned out that I not only liked it, I just loved it. Still do.
In case of my loudspeaker, it took us (my friend and myself) and a "jury" of around 20 people (not all at once, but in turn) to agree we had finally got it "voiced" just right all of six months, with at least three sessions every week. And frankly, I couldn't care less what somebody else thinks of it, because I'm happy with it.
There isn't a test in the world which cannot be rigged the way somebody wants it - may be hard, may be long, but it can be done. The best it can do is to produce a most probable finding, but some will still think differently. For any reasonable certainty, your statistical sample would have to be 1,024 people.
Anyone done such a test? And more importantly, did YOU agree with the general finding?
My problem is usually not hearing things that I want to hear, like controversial line conditioners. And this is often in the presence of people who say they can hear them. So of course the assumption is I don't hear very well.
However, when there is something obviously wrong I often hear it and diagnose it before a lot of people do. One case in point: a unit shipped for evaulation of a wireless linked subwoofer had a large latency that was uncompensated in the accompanying soundbar system. I said almost immediately, Something is terribly wrong with the integration of the sub and the rest of the system.
The other person presumed it was EQ. I was doubtful. He measured things acoustically, couldn't see much. Fiddled with the EQ. Didn't help. Finally the manufacturer of the wireless link is contacted. Given the part number.
"What are you doing with THAT??" That's a new multichannel module that is not released yet! It has a fifty millisecond latency! <I began to mock-pound my head against the wall at this point in the conference call>
Yes, agreed, they do represent first-rate engineering. That's why they're so wildly successful. Hard work for the designer to make easy work for the user.
- Status
- Not open for further replies.