Preservation of the National Library of Australia's Oral History Collection
Introduction
As a consequence of technological change SPATS has been compelled to embrace digital audio. We are now making digital copies on CD-R and an analogue safety copy on reel of items in the National Library of Australia's Oral History collection.
- There are over 28,000 items in the collection, comprising nearly 30,000 hours, stereo and mono, of recorded material.
- The carriers are almost all tape based, acetate, PVC and polyester, with a variety of magnetic particles and binders. There are examples of material right across the 50 years of magnetic tape recording.
- Many of the tapes are old, unstable and deteriorating, all require access and safety copies. To deal with this SPATS preservation copies around 1,500 hours of recorded material per year. Until recently this was onto a stable analogue format.
- All items in the collection are unique, ie the National Library of Australia possesses the only copy and has a clear archival responsibility for these items.
- The major part of the collection is now acquired directly on Digital Audio Tape (DAT) @48kHz, 16 bit linear. Though excellent for acquisition and possessing very high reproduction quality at the time of recording, the DAT is suspect as a long term storage item.
- Commercial support for professional analogue tape and related equipment is beginning to wane.
- There is no single apparent successor for the analogue reel tape, and the plethora of digital formats and the continued shifting of standards is likely, ironically, to be the only stable factor.
The limitations of the standard analogue method for preserving audio material have always been apparent. It can only be carried out in real time, introduces noise and distortion some of which is irreversible, and the whole process will have to be repeated 50 years or so, when the archival carrier itself will be in danger of degradation. To preserve the Library's digital original (DAT) by analogue sequential copying would mean converting from digital to analogue, a process that introduces distortion, and recording onto a magnetic tape that introduces noise. This negates the advantage of digital audio. It is also unlikely that an analogue tape player will be commercially available in 50 years.
These factors have caused a shift in electronic archival thinking.
The sound archivist no longer focuses on extending the archival life of the
carrier as far as possible. Instead, the archivist aims to select a digital
standard that preserves the integrity of the recorded material, and a carrier
whose archival life will exceed its projected obsolescence. And as the
recopying period is now much shorter, the rate of the migration of the digital
audio, and the staffing costs to do it are the major issues.
Compact Disc, Recordable (CD-R)
CD-R is not the ultimate solution to digital archiving, many other technologies offer some specific advantages. Overall, however, it appears to be the broadly acceptable solution to a number of technical issues.
CD-R Advantages
- CD stores digital audio in an easily retrieval and readily useable form. Access times per disc are quite fast. PQ codes allows for fast access to specific information.
- In audio format (red book) there has been a massive market penetration, millions of CD players have been purchased. Consequently it is likely that the effects of technical change will be gradual, and format support will be available for a time after its eventual demise.
- Audio is stored on CD-R in a 16 bit, 44.1 kHz linear, a form which is technically acceptable and widely utilised.
- A CD can be incorporated into a jukebox and automated for unattended back up of data to other and future digital technologies. Jukeboxes will also provide for local online access.
- CD-R is a write once technology and cannot be erased or overwritten once the TOC (table of contents) has been written.
CD-R Disadvantages
- It is limited, in its present form, to a 16 bit, 44.1 kHz data stream.
- The write once technology, unlike magnetic technology which can be overwritten, means an easily made error in the audio/PQ processing has produced an unusable CD which must be discarded.
- The organic dye is subject to degradation (as are other components of the CD).
- It is heavily dependent on its inbuilt error correction to make it playable.
- Data transfer for CD-DA is limited by the player's capacity to around 4x.
To minimise the failures associated with writing CD-R's, to verify levels, and to ensure that the PQ codes are placed correctly, the audio material is first assembled on a hard disc based digital audio workstation (DAW). The CD is then written on a SCSI writer in the background while the operator starts recording and assembling material for the next CD-R.
Every CD-R made will be checked on a CD tester and the relevant data relating to each disc stored. Longer term testing of the collection will then allow tracking of the degradation of the carriers. A decision has yet to be made about the type of tester.
Manufactured and Recordable CDs
The manufactured CD in all its formats is a different piece of technology to the CD-R. The CD-R is designed to be played on a standard CD player, but in a way that is entirely unlike the manufactured CD. A CD-R operates by virtue of a layer of dye over a reflective layer of gold. The laser distorts or melts the dye layer so that a later player is fooled into reading it as a series of "pits and lands", like those found pressed into the manufactured CD. Very little of the data derived from the life testing of a manufactured CD can be applied to a CD-R as the chemical make up and expected failure mechanisms are most likely to be entirely different.
CD-Rs ain't CD-Rs!
There are two types of CD-Rs available. Any discussion of their relative merits is coloured by the fact that much of the data comes from the manufacturers themselves, and they all hotly deny what is said about them. Evidence of this discussion can be found on the Kodak and TDK sites:
Kodak
TDK
- TDK Recordable Compact Disks
- Introducing TDK Recordable Compact Disks
- TDK Recordable Compact Disks - Questions & Answers
- TDK Recordable Compact Disks - Specifications
Cyanine
Discs: the first recordable CDs available used this type of dye. It has a short
unrecorded shelf life of about two years, and unsubstantiated tests set its
readable life at about 2/3rds of its competitor. Early CD writers, which wrote
only in real time, had a lower laser power than those now available, and
Cyanine dye may produce a recording with a lower error rate than its competitor
when coupled with these machines, than when recorded on the higher laser
powered higher speed writers.
Phthalocyanine
Discs: This, the second generation of recordable CDs, should have an unrecorded
shelf life of five years, and if the tests are to be believed, an archival life
well in excess of the projected obsolescence of CD technology. The higher
powered laser used in the most recent, and faster, CD recorders means that not
only is the dye modified, but the polycarbonate layer is also distorted when
recording on the phtyalocyanine discs. Though comprehensive tests have yet to
be carried out, this could mean that it is possible to retrieve data from a
CD-R even after the dye has failed.
Cyanine discs are a deeper green colour, and the phthalocyanine are a more gold colour, when viewed from underneath. However, the debate on reliability and archival life should not be reduced to these polarised viewpoints. Manufacturers are constantly monitoring and improving on their dyes, and the additives that are included in the discs' make up can modify some of the shortcomings of the original dye types.
Error Testing, Failure Mechanisms and Life Expectancy
The tentative, and wildly varying, estimates of life expectancy of the CD-R has its roots in a number of issues:
- The difficulty in measuring the Block Error Rate (BLER). The mechanism that feeds the data to the error 'compiler' are often no better than a CD player and error measurements are consequently difficult to repeat and may reflect the machine's errors rather than the disc's.
- Difficulty in identifying the failure mechanism. As CD-Rs are not very old technological devices, the way they will breakdown naturally is not known. Artificial age testing identifies a likely mechanism and attempts to accelerate that process. Failure to identify the correct method of aging and breakdown negates the tests. An egg, artificially aged by heat and moisture increase, is probably good to eat, while a naturally aged one is an entirely different item.
- Failure to achieve definitive results. The artificial aging processes have not produced the catastrophic failure BLER(max) needed to define end of life. Instead, estimates have been based on a lower BLER, termed middle of life, and results have been extrapolated from this.
The archivist needs to measure the increases in error rate in the
collection holdings as they age under normal conditions. This will lead to a
predictable useful life of the material held on CD-R, ensure an adequate
recopying or transfer period and maintain the integrity of the material
recorded.
Other Reading
MIX Magazine (July 1995) / Steven St. Croix and MIX Magazine (August 1995) response to the July article from TDK
Task Group IV, American National Standards Institute / Audio Engineering Society joint Technical Commission IT9-5 on Permanence of Magnetic and Optical Systems.
Kevin Bradley can be contacted at:
Sound Preservation and Technical Services, National Library of
Australia
Canberra ACT 2600
Telephone: (06) 2621655
Facsimile: (06)
2571703
E-mail: kbradley@nla.gov.au