| "Read Until Right" RUR - Part 1 |
The concept of using rereading of an optical, digital audio
disc ad infinitum, until a certain percentage of bits are
recovered, is not new. Powerful workstations in music
studios, like the "Plextor SADIE" do NOT need to employ any
error correction, but rather, they read and reread the CD to
verify that the data on the CD matches the actual studio
master tape. And, there are important reasons for this radical
departure from the norm:
disc ad infinitum, until a certain percentage of bits are
recovered, is not new. Powerful workstations in music
studios, like the "Plextor SADIE" do NOT need to employ any
error correction, but rather, they read and reread the CD to
verify that the data on the CD matches the actual studio
master tape. And, there are important reasons for this radical
departure from the norm:
| Error Correction creates Jitter |
The widespread belief that error correction in CD audio is
"perfect" has led to a complacency in digital reading and
design. The creative lunatic fringe applies inexplicable
tweaks, and observes important improvements, but they
are cast aside as if ECC and parity itself has accomplished
perfection. The myth of perfection would lead us to relax
and seek no more this passion that had captured all of our
creative energies in the analog era.
It is important to distill ECC, parity and jitter down to their
actual definitions to understand what they accomplish,
and more importantly, what is the cost of what they
accomplish. To understand jitter is to understand all things
gone wrong in digital audio. So let's begin.
"perfect" has led to a complacency in digital reading and
design. The creative lunatic fringe applies inexplicable
tweaks, and observes important improvements, but they
are cast aside as if ECC and parity itself has accomplished
perfection. The myth of perfection would lead us to relax
and seek no more this passion that had captured all of our
creative energies in the analog era.
It is important to distill ECC, parity and jitter down to their
actual definitions to understand what they accomplish,
and more importantly, what is the cost of what they
accomplish. To understand jitter is to understand all things
gone wrong in digital audio. So let's begin.
When an error occurs in reading a CD, a variety of techniques can be actuated to restore the read file to its original state. While
these error corrective measure do indeed correct, they create byproducts. Reed-Solomon codes (ECC) are quoted and misquoted
ad nauseum. Almost super-human attributes are attributed to these codes, without any genuine knowledge of what they can and
simply cannot do. The ECC chosen to correct errors in audio CDs is known as RS255/223 (see more details in our technical paper).
It is NOT perfect, it is NOT instantaneous, and it is NOT able to replace the misread (dropped) bits to exactly the point where they
were on the original. The reason? Time. In CD-speak, time is JITTER.
So what is jitter? Jitter is a misread bit within a defined moment of time. In analog terms, it most closely matches phase
errors. The presence of jitter indicates that all components of the data are not arriving (reading) in the original temporal format. It is
well known that the Reed-Solomon code chosen for MUSIC CDs is vastly inferior to the code used for PROGRAM CDs. The
reason? Real estate. Up to 50% of the area on a CD can be used for error correction on program CDs. On music CDs, the tolerance
for errors is relaxed to gain 15%-30% more area for music files.
There are several error corrective devices used in concert on a CD. Parity Bits are a checksum device that measures the whole
packet (called a "block") and adds or subtracts a bit if the sum in the block is incorrect. ECC (RS255/223) has a resolution of 1BIT.
This means that it cannot ascertain ANY information about the bit, only that it existed or not. In the sense of space and time,
transferred into analog terms, a bit representing a given frequency with phase shift of 0* or 179* will have the same output!
these error corrective measure do indeed correct, they create byproducts. Reed-Solomon codes (ECC) are quoted and misquoted
ad nauseum. Almost super-human attributes are attributed to these codes, without any genuine knowledge of what they can and
simply cannot do. The ECC chosen to correct errors in audio CDs is known as RS255/223 (see more details in our technical paper).
It is NOT perfect, it is NOT instantaneous, and it is NOT able to replace the misread (dropped) bits to exactly the point where they
were on the original. The reason? Time. In CD-speak, time is JITTER.
So what is jitter? Jitter is a misread bit within a defined moment of time. In analog terms, it most closely matches phase
errors. The presence of jitter indicates that all components of the data are not arriving (reading) in the original temporal format. It is
well known that the Reed-Solomon code chosen for MUSIC CDs is vastly inferior to the code used for PROGRAM CDs. The
reason? Real estate. Up to 50% of the area on a CD can be used for error correction on program CDs. On music CDs, the tolerance
for errors is relaxed to gain 15%-30% more area for music files.
There are several error corrective devices used in concert on a CD. Parity Bits are a checksum device that measures the whole
packet (called a "block") and adds or subtracts a bit if the sum in the block is incorrect. ECC (RS255/223) has a resolution of 1BIT.
This means that it cannot ascertain ANY information about the bit, only that it existed or not. In the sense of space and time,
transferred into analog terms, a bit representing a given frequency with phase shift of 0* or 179* will have the same output!
The alternative error correction method to ECC and Parity Bits is Read Until Right or RUR.
RUR is a powerful technique where a music CD is read at very high speeds until a bit drop is
detected. When a bit is dropped, the location is programmed to be REREAD. This continues
throughout the reading of the disc until as many as 99 rereads have been made to capture the
dropped bit! The obvious benefit is that it is actual data. Not a reconstruction code. RUR is
sometimes referred to as "SmartSampling" as it resembles upsampling with "intelligence".
Upsampling is "dumb" in that it "fires" a given amount of times at the same bit (4X, 8X,etc) so
more 'opportunities' to read that bit are had. RUR is far more sophisticated than
Upsampling or a relattively unsophisticated freeware CD ripper (in fact, RUR is NOT
technically a "ripper", sorry, no more information until we get our copyright) "knows" if a bit was
read and spends no time on captured bits. It does, however, reread a dropped bit not only
99X, but adjusts the laser to try to capture the bit. Some rereads will start reading early, some will
attempt a later initiation, some rereads will overlap entire sections (blocks and sectors) to
properly read the dropped bit, as most bit loss is due to the inability to precisely ascertain the
initiation and termination points of a block of bits.
RUR is the lowest jitter form of CD reading known.
Worth repeating, professional studio software and gear use rereading, and NOT
error correction codes, because of the jitter these codes create.
RUR is a powerful technique where a music CD is read at very high speeds until a bit drop is
detected. When a bit is dropped, the location is programmed to be REREAD. This continues
throughout the reading of the disc until as many as 99 rereads have been made to capture the
dropped bit! The obvious benefit is that it is actual data. Not a reconstruction code. RUR is
sometimes referred to as "SmartSampling" as it resembles upsampling with "intelligence".
Upsampling is "dumb" in that it "fires" a given amount of times at the same bit (4X, 8X,etc) so
more 'opportunities' to read that bit are had. RUR is far more sophisticated than
Upsampling or a relattively unsophisticated freeware CD ripper (in fact, RUR is NOT
technically a "ripper", sorry, no more information until we get our copyright) "knows" if a bit was
read and spends no time on captured bits. It does, however, reread a dropped bit not only
99X, but adjusts the laser to try to capture the bit. Some rereads will start reading early, some will
attempt a later initiation, some rereads will overlap entire sections (blocks and sectors) to
properly read the dropped bit, as most bit loss is due to the inability to precisely ascertain the
initiation and termination points of a block of bits.
RUR is the lowest jitter form of CD reading known.
Worth repeating, professional studio software and gear use rereading, and NOT
error correction codes, because of the jitter these codes create.
It is not well known to what degree noise corrupts CD error correction; HOWEVER, noise has NO effect on RUR, making it
immune to errors of that nature as well.
But rereading and amelioration of error code jitter are only half of why RUR has such low jitter. As RUR is rereading, the CD data must
be "stored" somewhere while the rereading takes place. The data is written to ELECTRONIC memory (no hard drives). The reason to
use non-mechanical memory is that hard drives, being spinning mechanical devices, create their own jitter, and it would be added to the
pristine data on the memory.
So we now find that after 99 rereads, we have almost all musical information on the CD, on the memory and it is virtually jitter free and
uses NO error correction codes. Pristine.
To that end, how do we................................................. "Play It From Memory?"
immune to errors of that nature as well.
But rereading and amelioration of error code jitter are only half of why RUR has such low jitter. As RUR is rereading, the CD data must
be "stored" somewhere while the rereading takes place. The data is written to ELECTRONIC memory (no hard drives). The reason to
use non-mechanical memory is that hard drives, being spinning mechanical devices, create their own jitter, and it would be added to the
pristine data on the memory.
So we now find that after 99 rereads, we have almost all musical information on the CD, on the memory and it is virtually jitter free and
uses NO error correction codes. Pristine.
To that end, how do we................................................. "Play It From Memory?"
Most critics of RUR simply recite that the invocation
of ECC is a rare occurrence, and therefore has no
need for improvement. If this is so, then ECC must
not offer much correction true? NO.
Measured results and subjective tests would indicate
that neither extreme is true. Parity Bit correction is
frequent, and suffers from similar jitter problems as
ECC.
The most common event that "fools" the
Reed-Solomon based corrective devices is
NOISE.
The noise of the chips and transistors themselves is
thermal and sub-atomic. Additionally, there is noise
from the operation of the devices. Noise can even
"fool" corrective measures into assigning a "1" or "0"
where none was present.
The category of noise that disrupts ECC and Parity
the greatest is known as Gaussian Noise because
of its predictable probability of distribution. or density
(POV).
of ECC is a rare occurrence, and therefore has no
need for improvement. If this is so, then ECC must
not offer much correction true? NO.
Measured results and subjective tests would indicate
that neither extreme is true. Parity Bit correction is
frequent, and suffers from similar jitter problems as
ECC.
The most common event that "fools" the
Reed-Solomon based corrective devices is
NOISE.
The noise of the chips and transistors themselves is
thermal and sub-atomic. Additionally, there is noise
from the operation of the devices. Noise can even
"fool" corrective measures into assigning a "1" or "0"
where none was present.
The category of noise that disrupts ECC and Parity
the greatest is known as Gaussian Noise because
of its predictable probability of distribution. or density
(POV).
| End of Section |
| The original marquee of the play "RUR" by Karel Capek in Prague, 1920. |
| If the proper mathematics are applied here (to the polynomial), there is a 99.5% chance of some jitter being created in every bit read mechanically using ECC and Parity Bits correction. |
| Jitter in Space/Time |
 |
R.U.R. was written in 1920, premiered in
Prague early in 1921, and was performed in
New York in 1922. Virtually every
encyclopedia or textbook etymology of the
word "robot" mentions the play R.U.R. by
Karel Capek
Prague early in 1921, and was performed in
New York in 1922. Virtually every
encyclopedia or textbook etymology of the
word "robot" mentions the play R.U.R. by
Karel Capek
| Alternative methods to extract musical information from a music CD WITHOUT USING ERROR CORRECTION |
"Dumb" & "Smart" technologies, Upsampling, Selectable Algorithms, Reclocking, Clock-linking,
Clock Synchronization, computer based extractors and many more. ALL of them fail in one
imperative arena of thought: TIME. That ALL have no ability to ascertain WHEN the bit was
written. In the beginning of the 44100hz sample (jittered)? The end (jittered)? Dead Center (no
jitter)? The ONLY TRUTH lies stamped on the disc as it was stamped by the master.
So instead of trying to correct the errors created by error correction, clocks syncing to each other
& worst, reclocking, why not use the ORIGINAL DATA...read on :
Clock Synchronization, computer based extractors and many more. ALL of them fail in one
imperative arena of thought: TIME. That ALL have no ability to ascertain WHEN the bit was
written. In the beginning of the 44100hz sample (jittered)? The end (jittered)? Dead Center (no
jitter)? The ONLY TRUTH lies stamped on the disc as it was stamped by the master.
So instead of trying to correct the errors created by error correction, clocks syncing to each other
& worst, reclocking, why not use the ORIGINAL DATA...read on :
| GAUSSIAN... NOISE |