Perhaps about the same number of denials of a true statement to turn it into falsehood?Jakob2 said:
yes.... this is what i am wondering about... if any possibility or change occures here. Bit rounding... anything. And, are different systems handled better or worse than some others. Proof, pls.
THx-RNMarsh
Coincidentally, that's exactly what I use these days for playing CDs, as well as WAV/FLAC files. My faithful CD player is in the basement, collecting dust.
But as noted before, software distribution is a similar problem and that has been honed to near-perfection. As I said, when was the last time your username was misspelled on diyaudio.com? 😉
The preparation process is not perfect (nothing ever is, except Scarlett) but it should be orders of magnitude (many orders of magnitude) better than the clunking CD mechanical stuff and laser high-frequency decoding, and that is already pretty faultless. I believe your worries would be better aimed at other real problems. 😎
Jan
Mark, you just stepped into a deep puddle of popularity with me. Buy a blue ray drive to upload your cd's and everything else on shiny platters to a server and be happy forever.
I notified Chris that I quoted his post, in case he wished to enter the discussion if questions were raised. He's busy with other things at the moment.
Last edited:
Just asking..... up to and just prior to encoding. BTW - username change is not the same as music in that we can corrupt music a little and it is still listenable/usable.
Ok. Never mind.
-RNM
The biggest change in the CD sound - for me - has been thru changing to loudspeakers which do not compress the dynamic range to any degree I would subject myself to.
In listening to a tune today, I got out the SPL meter to see what I was listening to.... I thought I was playing loudly enough.... 75-85dB SPL average at listening location but with peaks captured to 95-98dB -- and from a cymbal crash, no less. On a rock song! Apparently this has been my major malfunction in CD playback systems... dynamic range and speaker compression (and distortion).
cymbal crash peak spl at listening location.
Now, I will have to go thru the CD's to find ones with greater dynamic range than this one.
THx-RNMarsh
In listening to a tune today, I got out the SPL meter to see what I was listening to.... I thought I was playing loudly enough.... 75-85dB SPL average at listening location but with peaks captured to 95-98dB -- and from a cymbal crash, no less. On a rock song! Apparently this has been my major malfunction in CD playback systems... dynamic range and speaker compression (and distortion).
cymbal crash peak spl at listening location.
Now, I will have to go thru the CD's to find ones with greater dynamic range than this one.
THx-RNMarsh
Last edited:
Easily done, but no one really wants to take the time. Talking about it and guessing is better.
Hi Everyone, DF96,
I see a quote has someone unsure of what I meant in it. The quote:
You`ll have to excuse my use of the terms C1 and C2 error flags. These are what was taught to us at the beginning of the CD age. These used to be available as test points that we could monitor. DAT and other similar technologies used the same concepts and `flags`. So, to be clear on this simple concept, the C1 and C2 flags. There is a built in expected number of errors per revolution, called (at that time) BLER. The BLock Error Rate. I forget what the normal range was for that. These would activate the C1 flag and thus be counted while another signal indicated the beginning or end of the block. The counter was reset on the block signal. I`m a little rusty on this as it has been a couple years in the past. If you have to take something away from what I said in the quote, it is that the C1 flag indicated a data error had been detected. Nothing more. Most of these errors are corrected using data that was interleaved in the surrounding data. The C2 error flag. That`s the bad one. The C2 flag means that the data error was not able to be corrected (normally due to surrounding bad data). The C2 flag indicates that the data cannot be reconstructed, period. It marks that frame of data bad and continues processing detected errors.
When a bad packet of data is received by the next stage, a number of possibilities exist depending on how good that CD player is. The cheap ones make the data valid (not correct, it just satisfies the CRC) and pass it down to the DA converter. Now you know why some machines are awful sounding. The other options are mute, copy the previous data (same value), or in the better machines, interpolation where the error correction looks at the previous and following data values and uses a value in between those. Obviously, these machines sound better, although in the case of only one error, you wouldn`t hear the mute to begin with. Oversampling the data generates more intermediate values, so a much smoother correction.
What does this all mean to our `perfect bit reproduction`folks. It disappoints the lot, because the only correction a C2 flag gets will not be the original data except by fluke. But, you are not getting a bit perfect reproduction as soon as the C2 flag goes up.
What about the lack of digital errors then. The correction makes the corrupted data obey the CRC correction by either replacing or muting the original data in a format that is valid data. It`s just not what the master has on it. When you hear disturbances in the music or test tone, you are hearing a string of bad data packets, not just one. One C2 error in a base sampling rate system would create a `glitch`lasting 1 over the sampling rate, or 1 over 44,100 sec long. You are unlikely to hear that short a disturbance.
Hi Jakob2,
The long and short of all this is simple. Enjoy the material on the CD and accept that it isn`t bit perfect. As for CDROM players, they can re-read a damaged section all they want. If they are using CD decoding rules (you are playing a music CD), they will be wrong almost as often as a cheap CD player. A good CD player probably has much better concealment and interpolation than the computer CDROM does. Only if you are listening to music made with CDROM data encryption can you hope for real bit perfect reproduction. These discs do not play on a CD player.
-Chris
I see a quote has someone unsure of what I meant in it. The quote:
Quoting the response ...
... and Mark points something out that is important ...
Now, of course Mark is correct here. The data disk and music disks use completely different encoding. Don`t confuse the two of them.
You`ll have to excuse my use of the terms C1 and C2 error flags. These are what was taught to us at the beginning of the CD age. These used to be available as test points that we could monitor. DAT and other similar technologies used the same concepts and `flags`. So, to be clear on this simple concept, the C1 and C2 flags. There is a built in expected number of errors per revolution, called (at that time) BLER. The BLock Error Rate. I forget what the normal range was for that. These would activate the C1 flag and thus be counted while another signal indicated the beginning or end of the block. The counter was reset on the block signal. I`m a little rusty on this as it has been a couple years in the past. If you have to take something away from what I said in the quote, it is that the C1 flag indicated a data error had been detected. Nothing more. Most of these errors are corrected using data that was interleaved in the surrounding data. The C2 error flag. That`s the bad one. The C2 flag means that the data error was not able to be corrected (normally due to surrounding bad data). The C2 flag indicates that the data cannot be reconstructed, period. It marks that frame of data bad and continues processing detected errors.
When a bad packet of data is received by the next stage, a number of possibilities exist depending on how good that CD player is. The cheap ones make the data valid (not correct, it just satisfies the CRC) and pass it down to the DA converter. Now you know why some machines are awful sounding. The other options are mute, copy the previous data (same value), or in the better machines, interpolation where the error correction looks at the previous and following data values and uses a value in between those. Obviously, these machines sound better, although in the case of only one error, you wouldn`t hear the mute to begin with. Oversampling the data generates more intermediate values, so a much smoother correction.
What does this all mean to our `perfect bit reproduction`folks. It disappoints the lot, because the only correction a C2 flag gets will not be the original data except by fluke. But, you are not getting a bit perfect reproduction as soon as the C2 flag goes up.
What about the lack of digital errors then. The correction makes the corrupted data obey the CRC correction by either replacing or muting the original data in a format that is valid data. It`s just not what the master has on it. When you hear disturbances in the music or test tone, you are hearing a string of bad data packets, not just one. One C2 error in a base sampling rate system would create a `glitch`lasting 1 over the sampling rate, or 1 over 44,100 sec long. You are unlikely to hear that short a disturbance.
Hi Jakob2,
Absolutely!
Nicely put.
The long and short of all this is simple. Enjoy the material on the CD and accept that it isn`t bit perfect. As for CDROM players, they can re-read a damaged section all they want. If they are using CD decoding rules (you are playing a music CD), they will be wrong almost as often as a cheap CD player. A good CD player probably has much better concealment and interpolation than the computer CDROM does. Only if you are listening to music made with CDROM data encryption can you hope for real bit perfect reproduction. These discs do not play on a CD player.
-Chris
Last edited:
If you monitor the Error Flag (what you're calling C2, I think) in a non-broken player, it just doesn't fire very often. Back when I had a Magnavox 582, the signal could be read (with a pulse stretcher!) from pin 4 of the 7220. At that point, I could be certain that the data delivered to the 1541 was bit-perfect. Stereophile and HFN/RR reported the same result.
Both C1 and C2 have three flags. To measure BLER, the system monitors and adds the C1 flags E1.1, E2.1 and E3.1.
Uncorrectables are monitored on the E3.2 flag of C2. E1.2 and E2.2 indicate that there were errors and that they were corrected.
Uncorrectables are monitored on the E3.2 flag of C2. E1.2 and E2.2 indicate that there were errors and that they were corrected.
Rich, your user name (or any other example) can be corrupted by just a single bit flip. And the data comes from the other side of the world, often in split packets that arrive via different pathways and are stitched together at the server farm. The fact that in all those years nobody called you, say, SNMarsh on diyaudio is a testament to the robustness and for practical purposes perfect digital processing and transportation.
Logically, it doesn't make much sense to worry about that part of the preparation of a CD source file in comparison with the CD reading/decoding process; even 1 uncorrectable error per day is vastly more bad than the digital front end side of things.
Jan
Last edited:
Hi Chris,
I don't think that is the case. All along these last pages I at least (others too) have made it clear that the C1 error corrections, which occur sometimes 1000's of times on a CD, produce perfect bits, and it appears you agree.
C2 errors can generally not be corrected and lead to wrong values being send to the DAC. That is all recognized.
The point is that C2 errors are rare and there are many disks that have zero C2 errors. Even if they have errors, it is generally immediately audible, but as a transient type of 'bad sound'.
It is not something that changes the sound between players in normal listening.
Jan
Jan, C2 is the second round of error correction.
http://cloversystems.com/products/disc-testing/dvx-4/
I posted this link so that there would not be an endless discussion on how error correction/detection works.
Fig.24 (page 39) clearly shows that there are errors not corrected by C1 (E3.1) that are then corrected by C2 (E1.2 and E2.2). All errors were corrected as E3.2 = 0.
E3.x flags are uncorrectable flags. The rest indicate that a correction took place.
http://cloversystems.com/products/disc-testing/dvx-4/
I posted this link so that there would not be an endless discussion on how error correction/detection works.
Fig.24 (page 39) clearly shows that there are errors not corrected by C1 (E3.1) that are then corrected by C2 (E1.2 and E2.2). All errors were corrected as E3.2 = 0.
E3.x flags are uncorrectable flags. The rest indicate that a correction took place.
Last edited:
I thought the CD standard (Red Book?) requires interpolation? If so, those players which do not interpolate are non-compliant and so should not be able to display the 'CD' logo. Sending known faulty data to the DAC is inexcusable - do some machines really do that?anatech said:
About once an hour, I think SY told us?
I stand corrected ;-). I merely wanted to make a distinction between 100% correctable errors, which abound on a typical CD, and the rare or absent uncorrectable errors.
Jan
Not quite that often, I had to go through several discs to get even a single error.
HFNRR back in the day even ran an article showing how to tap the flag (I think it was Atkinson) in the 1541/7220 chipset, which I immediately tried, as did many others. Everyone got the same result- it took significant disc damage to trigger it.
There was an Elektor kit that tapped the various flags and counted and displayed the different types of errors.
I build it at the time, and was horrified by a) the 1000's of corrected errors on a typical CD and b) the very rare occurrence of uncorrectable errors, most CDs zero.
Jan
I build it at the time, and was horrified by a) the 1000's of corrected errors on a typical CD and b) the very rare occurrence of uncorrectable errors, most CDs zero.
Jan
All this raises a sort of "Angels on the head of a pin" question.
Since I can rip 100s of CDs with something like dBPoweramp and get correct checksums and no reported errors - does this mean that the CD-ROM drive is doing something that an ordinary CD player is not? E.G., doing rereads to try fixing bad spots? Uncorrectable errors may be rare on CD players, but they are very rare on ripping software. Why? And does it even matter?
Since I can rip 100s of CDs with something like dBPoweramp and get correct checksums and no reported errors - does this mean that the CD-ROM drive is doing something that an ordinary CD player is not? E.G., doing rereads to try fixing bad spots? Uncorrectable errors may be rare on CD players, but they are very rare on ripping software. Why? And does it even matter?
- Status
- Not open for further replies.