2 August 2024, NAA Offices Perth
Managing Risk with AV archives – Ben Chesterman NAA
Three aspects that need to be thought about when considering risk for media preservation.
Physical media
NAA has a vinegar syndrome label for their cans, includes acid level and date. Also part of their metadata. Vinegar syndrome has OH&S considerations – can damage skin/ cause irritation.
Increasing opportunities for home grown gear to playback media in old formats.
Digitisation
Really important to emphasise to management that the digitisation step is only one small part in digital preservation!
Make sure the digitised files link back to the original physical master. Are our MFA links clear enough?
Cyber security compliance for old workstations is becoming a problem e.g. 10-yr old specialised equipment can’t be updated with latest patches.
NAA has born digital file standards for AV – have a quick reference. https://www.naa.gov.au/information-management/storing-and-preserving-information/preserving-information/born-digital-file-format-standards
Outsourcing – virus scanning on delivery under cybersec policies – big problem for LTO tape delivery. Lesson: need timeline flexibility in outsourced digitisation because things go wrong, and even managing the service provider relationship itself.
Quality checks: Ben, with AV manager experience, automated QC checkers are not bullet-proof – you need to put eyes on the digitised materials. QC metadata and reports must be kept alongside with the rest of the record for the object.
Digital preservation
Mediaflex system used by NAA – they trust the fixity of the storage used in this system. Putting out to LTO tapes, and a copy in Sydney and Canberra on Hitachi storage (private cloud?). NAA using MD5 checksums.
DROID not deep enough for AV metadata scraping, MediaInfo is better. Might be a problem for format obsolescence if you don’t know what you’ve got.
Video digitisation and DCWA – Deb Paisley & Jameson Feakes
EzDigital on Fitzgerald St are competent for personal archive digitisation according to Jameson.
DCWA is pitching for new equipment to scan 3D objects in their next funding round, likely to be housed at WA Museum. Bound book scanner is coming any day now.
DCWA piloting a metadata creation service for AV digitisation, video (automated?) delivering an amended version of the return spreadsheet they’re working with.
What’s their thinking on raw footage? esp if it hasn’t been produced? Thinking they will keep raw footage tapes because they have direct UWA relevance. Might not stick for born-digital materials? Auto-generated transcripts? Yes will be considering.
Should we add a transcript step to our DCWA returned analogue video material?
indicates the interaction between EDRMS Trim and Recollect. Recollect allowing inter-institution links between different collection objects held in various parts of the university e.g. Calloway Music Archives and University Archives.
Toolkit I – Jameson Feakes DCWA
For ingesting DV material from tape: use old PCs – adding Firewire cards to them later. Or finding old computers with FW sockets built in.
Jameson has learnt Python as glue for his automation efforts. About time I had a crack at learning Python?
Toolkit II – Ben Chesterman NAA
There is a difference between different film scanners in quality / perception. e.g. ScanStation vs Cintel these will work differently for different stocks and different film gauges.
Cyan and Yellow dyes fade first in film, hence some poorly stored old film may have a magenta tint, which can be corrected from the DPX’s.
If outsourcing digitisation, write scripts to QC metadata of incoming digitised files using MediaInfo to compare against our expected requirements? e.g. format validation, confirming tools/equipment used.
Teracopy: MD5 and SHA256 are too slow when compared to xxHash options, NAA is considering adopting this format for their fixity checking as their standard is SHA512. xxHash is developed by ex-Googler. James Doig is exploring this at the NAA.
NAA using 7Zip to create their TAR files.
Best place to start for web archiving research is IIPC WARC tools project. Oral histories post production from two mic recording – try noise gating channels separately and merging together after – only on access file, not pres file.
AI: implications for archival methods and case studies – Piers Higgs & Meg Travers Gaia Resources
AI doesn’t necessarily fit into an OAIS model, but can provide extra features in different phases: e.g. ingest, access, administration
Archivists are good at trust and governance, this will be in good stead when considering how AI can be implemented. AI might speed up the time to access and discovery if it’s part of the workflow?
General horror in the room if source materials are OCR’d or handwriting recognition to text, and with language modernisation to make the content accessible to a modern audience. e.g. translating the ‘convict slang’ used in the early WA colony police occurrences books. Concerned that our archive users will never go to the source and read it for themselves. Gaia Resources tools present the original material alongside the OCR text so the computer output can be checked.
Q&A panel
Language models, community built with ‘ethically sourced’ training data – these might be the ones to look out for.
Leave a Reply