43rd Annual Meeting- Electronic Media Group Session, May 16, "The Fragile Surface: Preserving the CD-DA by John Passmore

John Passmore works as the archives manager at WNYC. They have audio recordings that go back to the earliest days of radioРabout 100 years now. As a listener of WNYC it was very interesting to hear about how they are caring for their archives. WNYC does a lot more than just radio- like many media producers, they are a multi-platform production company with born-digital content. In the audio preservation lab they are able to digitize and preserve almost all media types, which is an anomaly in the public media world.
That being said- between 2000-2008 they created ~30,000 digital audio cds which were created and finalized by the engineers at the time of the broadcast. About 5 years ago they started to notice some problems with them being unplayable and hard to rip. It was possible to extract the .wav file, but it would sound terrible. Being worried about the health of these formats they created a workflow to migrate them in an efficient and responsible manner. Some have obvious manufacturing defects, but the scary thing is that usually you can’t see something visibly wrong. The most commonly believed reason that cds fail is due to degradation or failure of the dye layer. This is extremely small- at 0.5 micrometers this is about 1/100th of the width of a human hair!
To migrate their collection, they bought a machine that is used to put cds onto an ipod or other digital player- it is not preservation oriented, but can concatenate .wav files dump them onto the DAM and extract the metadata. If the cd fails then they use a separate testing system. Plexstore can be used to review the data. It is also possible to run the cd in real time in a cd player and extract the info that way, but this is obviously much more time consuming. There are problems with this system and John is not totally happy with it, but it is working for now.
Now for the alarming information- they ran a test of about 20% of 2,400 discs to determine how many errors and what kind there were- correctable or uncorrectable. None of the cds passed! So the whole collection is at great risk.
John’s list of takeaways:

  1. CDs don’t last very long, maybe even less than we though- their CDs are only 10 years old, not the 20-30 usually stated in accelerated aging tests
  2. It is hard to know why the CDs go bad, and finally
  3. They are looking at open source tools like QCTools. John ended with a great video using MakeAGIF.com showing the process of his machine in process.