The Complete
Technical Analysis
Report on
Memory Playback
and RUR

Since the advent of digital audio on optical media with the introduction of the CD in 1982,
music listeners and designers in the audiophile community have faced continual challenges
in attempts to improve the quality of music reproduction.  The primary culprit has been
identified as “jitter”, which refers to errors in the timing of when the laser sees the sample
and begins to read for bits.  Also, we use the term “Bit Drift” to describe an error in timing
when reading the bits themselves.  

In this white paper, we will revisit the traditional practice of reading digital optical media
(with emphasis on 16 bit audio media).  We will introduce our process, called Memory
Playback, and propose how it will accomplish the following:

  • Eliminate error correction and demonstrate a flaw in the original code
  • Remove Random Access Memory (RAM and caches) to eliminate odd harmonics they
    may create in the DAC
  • Use only Solid Memory to play only music bits in their original order
  • Preserve inter-relative bit timing that is randomly lost in all optical media
  • Raise 16/44 performance to “analog” levels required for High End Audio
  • Provide the popular convenience of a stand-alone music server that plays original
    master quality CD extractions, with nearly unlimited support for Hi-Rez in the future

Behind the issue of distortion in the sound of the CD lies the accepted standard for mass-
media error correction, known as Reed-Solomon CIRC (Cross Interleaved Reed-Solomon
Coding).(1)   Its universal application from the first production of CDs resulted in the
ongoing assumption on the part of the listener that its process yielded no errors at all, in
either reading or playback, and was included under the umbrella term “bit perfect”.(2)

The subject of this paper, Memory Playback, eliminates the primary difference between
analog and digital audio, which is error correction. With error correction bits removed,
every bit is a representation of a moment of music, and nothing more. Error correction is
the practice of adding bits to conceal mistakes or replace dropped bits, but often at the
wrong time or sequence. Many contend it doesn't matter, as the system is perfect, or this
known fault, “is inaudible”.  What this claim actually means, and what “bit perfect”
accomplishes, will be explained within these pages.

END OF SUMMARY

Excerpt from Part One:  Reading

What Humans Can Hear:  The Problem of Bit Order

In 1982, N.V. Philips and the Sony Corporation released the standard for CD digital audio
systems.  It was designed to have the ability to detect and correct burst and random errors.  
The correction code is known as the CIRC, or Cross Interleaved Reed-Solomon Code.(1)  

Reed-Solomon encoding belongs to the family of burst coding, and was never intended for
sub-byte level integrity. Burst coding is defined as “byte level correction”.  The most
obvious outcome of any byte level error correction is the anomaly, or a series of such,
based on certain assumptions about the limitations of human hearing.  

The first assumption is that of the actual nature of the output, actual in the sense of time,
that is, the temporal integrity of contents of the byte.  This paper will explain briefly why
Reed-Solomon encoding is blind to bit-level data, and therefore incapable of Bit Integrity
(BI), as are all burst codes, the family to which Reed-Solomon encoding belongs.  In
addition, further assumptions take place based upon the limits of human perception being
temporally restricted to that which transpires during any given 25us period or less.

It was this decision, in fact, that was responsible for the error corrective devices to create,
once decoded to analog, complex and unrelated harmonic content based exclusively upon
the use of random, temporal reads in the sub-byte level domain. We contend that it is these
in total that are responsible for, and manifest as, the commonly used term for all things
abhorrent in digital audio: “digititis”.  The problem manifests as a sonic flaw, due to the
fact that error correction is blind to the actual location and order of the bits.

Error correction is the replacement of lost data through artificial or “redundant” data
targeted at the byte level.  The more bits that can be passed through any system, the faster
that system is deemed. Reed-Solomon error correction cannot actually correct individual
bits, but rather, it can only replace groups of bits known as “bytes”, “frames” or “blocks”;
all three terms being interchangeable and referring to the same thing.  When Reed-Solomon
replaces dropped bits (no data), it replaces a grouping of bits with no real ‘knowledge’ of
where the bits were (in time), or what order they were in.

It only replaces the quantity of bits.  This is to say that it can replace a byte (a group of 8
bits) without ascertaining any information about the bits. The size of these groups is
predetermined by the minimum “resolution” required by the system.

FOOTNOTES:
(1) Vries, L.B, Odaka, K., CIRC-The Error-Correcting Code for the Compact Disc Digital Audio System, AES Conference 1st
International Conference Digital Audio, June 1982
(2) See DIYAudio Forums>Top>Source>Digital>What Is Bit Perfect?, May 7, 2005

END OF EXCERPT



Thank you for your interest in The Memory Player.


Beyond Bit Perfect:  
Advancing the Art of CD Reading and Playback