One of the goals for our current work on the IMPAC start-up project is to test proposed preservation workflows. Fortunately, here at Indiana University, we aren’t starting with a blank canvas.
As we argued in Meeting the Challenge of Media Preservation, not only do we have valuable material on the IU Bloomington campus at risk, but we also have expertise and experience we can leverage to do something about this challenge. Much of our experience was shaped in service of multiple preservation projects. These projects produce much more than the preservation of valuable materials – they also forge strong partnerships between campus units, create tools, and stimulate research into methodologies and workflows that we can use, improve, and adapt. In other words, when used to their full potential, these projects are stepping stones – each one positions us for what comes next.
Just so with AHEYM, The Archives of Historical and Ethnographic Yiddish Memories project (the acronym means “homeward” in Yiddish). AHEYM is a digital video collection of approximately 800 hours of language interviews with 350 Yiddish speakers collected by IU professors Dov-Ber Kerler and Jeffrey Veidlinger along with their colleagues Dovid Katz and Moisei Lemster during ten fieldwork expeditions conducted between 2002 and 2009. The materials consist primarily of ethnographic, linguistic, and oral history interviews with Yiddish speakers, many of whom are Holocaust survivors. The recordings also include musical performances, folklore, and footage of contemporary Jewish life in Eastern Europe as well as historical sites of Jewish memory in locales in the region.
The AHEYM collection is being accessioned into the Archives of Traditional Music (ATM) at IU and will ultimately be accessible online. Here is a sample:
Most of the collection was recorded onto MiniDV tapes and you might think that this relatively recent digital format is easy to preserve. There are two basic approaches: a transcoding approach in which an audiovisual signal is captured complete with whatever error correction was necessary or a ‘native’ approach (sometimes called bit scraping) in which a bit-for-bit data stream is captured. There are significant advantages to the latter approach, including capture of embedded metadata that identifies errors and enables an assessment of the integrity of the transfer. This better supports preservation principles. It also results in a more efficient workflow that takes advantage of software tools and automated data analysis.
Simple, right? Just copy the data. But it’s not quite that simple: MiniDV is a finicky consumer level digital format and the data is contained on a physical (“analog,” in a sense) surface—magnetic tape. Nearly every transfer will result in a slightly different set of errors. So how do you design a preservation workflow that addresses this characteristic of the medium?
To our knowledge, few models for DV tape preservation workflows exist. Building upon prior personnel and video experience from the EVIA Digital Archive project, incorporating principles and best practices from Sound Directions at the ATM, and consulting with AVPS (AudioVisual Preservation Solutions), we built a robust workflow around free, open source, and/or low-cost software tools allowing us to automatically generate quality control data for every frame of captured video without an attended (monitored) transfer. And by putting this into practice and evaluating the results, we came up with:
1. A system of red flags we could use to suggest alternative preservation strategies for particular problematic tapes
2. A means for evaluating an overall sense of transfer “performance”
3. A method for identifying error regions based on these criteria: a) Are they detectable at playback speed and full frame size? b) Do they obfuscate significant content, and, of course, c) Are they present on the source?
Right now, we have finished processing close to 600 of the 800 tapes. For more information on our DV preservation workflow, please email Anthony Guest-Scott at firstname.lastname@example.org.
// Anthony Guest-Scott