AES Analog Preservation Panel
October 10, 2014, Audio Engineering Society Convention, Los Angeles—The problems of saving analog audio content is a big issue for researchers, historians, and anthropologists. Konrad Strauss from McGill University moderated this panel with panelists including Mark Hood from Indiana University, Nadja Wallaszkovits from the Austrian Academy of Science, Brad McCoy from the Library of Congress, and George Massenburg from McGill University.
McCoy opened with updates on the National Preservation Plan. This act of Congress created a oversight board, a study for action plans, and an action plan. The study, along with clir.org identified best practices and other information. Other foundations added series of actions to preserve, identify, and research the areas.
Strauss added that the Association for Recorded Sound Collectors (ARSC) added sections on scholarships, collections, back-end metadata, and other areas. They also added their scholarship about musicians and music.
Wallaszkovits commented that her organization is the European version of the US ones. www.iasa-web.org has many resources and publications on the preservation of audio visual materials and guidelines for the content in the organization's handbook TC04. another group, Open Archive Information Systems (OAIS) details processes for ingest, digitizing, adding metadata, and repositories. They suggest using standard workflows to produce 24-bit, 48 kHz .wav files. Another resource is at www.jazzpoparkisto.net/audio
Strauss stated some issues that the industry has to address. Ongoing media degradation means that we have a maximum of 15-20 years to convert all of the older materials. A better way to state this is that we have until 2029 to finish the job, since the 15-20 years statement has been expressed for over 20 years. We need to migrate the old analog audio content to digital to preserve it.
Massenburg suggested that companies could get greater incentives if they could see an easy way to monetize the content. A number of the 1/4th-inch archives have been lost due to the carrier medium decomposition. The return of high-resolution audio offers an opportunity to resurrect the analog masters.
McCoy commented on the importance of the materials versus their historical heritage. The Library has collections of over 1 M movies and 3.5 M audio sources but needs to have better public-private partnerships to continue their efforts. More companies could donate their masters to the Library, but this might change the relationships with the studios. The Library needs high-resolution originals and metadata.
The National Jukebox now holds pre1925 items from Victor, Columbia, and Edison including the discography using the original Victor logs. Other audio files of historical interest include acoustic recordings from 1900-1925, recordings of radio interviews of historical figures, original cylinder recordings, the first field recordings from a 1890 researcher capturing native American songs, and some vertical-cut lacquer records from '47.
Wallaszkovits noted that the creation and cataloging of metadata is a major horror story because it is so time consuming. The many archives have large amounts of content starting in 1899, and all of the original materials need metadata to further research. An exchange mechanism is key to access and interoperability across collections. The mapping across the many formats can be done online through multiple search criteria and linked to the online archives. There are some rights issues to be resolved for some of the content.
Massenburg noted that archiving and preservation can be done on a smaller scale to preserve everything without grand infrastructure. In many cases, it is not easy to go to the cloud, but there is still a need for reliable short-and long-term storage with risk mitigation, and manageable security for multiple levels of access. A general backup that is bundled with metadata and special production backup can be transferred to tape. The copies are verified and put on multiple servers.
Typical hardware could be a PC with dual SATA docks for networked storage, and archive copies to LTO-6. this storage array now holds 48 TB and all of the equipment cost about $10k, plus some for tape. A higher-level system could be assembled for $12-15 k. The hard disks are the limiting factor because they have to be redundant and geographically dispersed.
Since the justification for the archive hardware is based on an asset acquisition base, $10k is not excessive?
Hood responded that one part of the process is to assess the collection for archive or transfer. Perform an inventory to determine how many objects, the formats, and the amount of content ( channels x duration). Once digitizing has commenced, you need the services of a digital archivist and librarian.
You need to identify the technology resources and expertise on equipment for playback and storage. A big issue is the physical deterioration of the carrying media and the lack of equipment to read the content. The preservation requires a constant upgrade to physical and file formats, so you have to identify the formats to export to. If you don't have the resources, outsource the process since people with the expertise exist.
Standards and future retrieval are issues with all formats, so it is important to keep all equipment in the best possible shape. The transfer and archive process may be the last time this object will be played, but still try to save the original object. Digitize at the highest sample rate and at the deepest bit depth possible. Preserve any original metadata and remember that the clock is ticking.
Documenting legacy reporting techniques?
McCoy suggested capturing all information from a knowledgeable engineer, and try to save that domain expertise. Capture the expert knowledge from all in the field. Put the instructions on a video.
Hood added that anyone can contribute more basic how-to tapes and video tutorials. Nevertheless, you need to document all phases of the operations. Look at the ARSC resources.
Clipping and headroom?
McCoy noted that with Dolby you need tones, but most tones on tape are not good. It is important to set levels on content and not just rely on the tones.
At high resolution, 24-bits offer lots of dynamic range and headroom, 144 dB?
Hood commented that scratches and pops will still clip/
Standards for metadata, levels, and formats?
McCoy noted that AES 57, and 60 identify various parameters and other standards exist to help with data exchange. Multiple standards are ok, just think how to map the various standards to the archive.