In January 2002 I began a semester-long practicum at the LBJ Library and Museum working to reformat a collection of reel-to-reel audio tapes into digital files. I became interested in audio preservation during the course of my studies at the University of Texas at Austin's Conservation and Preservation Program. The practicum required research about current standards and reformatting procedures, the development of both standards and procedures to fit the needs of the reformatting project at the LBJ Library and Museum, and the implementation of these procedures by transferring some of the analog materials to digital files.
The LBJ Library and Museum was dedicated on May 22, 1971. It is part of a system of presidential libraries administered by the National Archives and Records Administrations. The library was established to preserve and make available for research the papers, photographs, audio and video materials, and memorabilia of LBJ's presidency. However, the library's holdings include materials from Johnson's entire public career, as well as materials related to his family and close associates. The library currently houses 40 million pages of historical documents. In addition, the library actively collects the papers of Johnson's contemporaries and conducts an oral history program.
The collection being reformatted includes the sound recordings of the president's speeches, press conferences, and public remarks known as the Presidential Speeches. There are eight hundred thirty-one reel-to-reel tapes in the collection with a total of twenty-one hundred speeches. The White House Communications Agency (WHAC) was responsible for most of the original recordings. When the speeches were original recorded, each one was put on its own individual reel. Each reel lasted approximately 30 minutes, so if the speech was longer than that, it was put on more than one reel. The length of speeches in this collection ranged from less than a minute to over an hour and a half. WHAC then combined different speeches of varying lengths to create reel-to-reel tapes with a combined length of approximately 30 minutes. The speeches were maintained in roughly chronological order based on how they fit together to form 30-minute reels. Thus 2,100 individually taped speeches were combined into an 831-tape collection known as the President's Speeches.
WHAC made many copies of the combined speeches including the master copy, the presidential copy, the working copy, the reference copy, and two archive copies. These copies are all currently stored on the third floor stacks at the library, an area that is kept cool and dry and monitored for both humidity and temperature. The lights in the stacks remain off unless someone is on the floor to retrieve an item. The original tapes are currently kept in the cold storage vault at the library and only used as a last resort. The president's copy is being used to make the transfers, with other copies being used only if necessary. A program of exercising the tapes every two years was begun in the early 1970s but this practice ended a decade later. The tapes have been stored tails out and overall are in very good condition.
The first assignment during my practicum was to create a digital finding aid to improve accessibility and establish better archival control. I designed and populated a Microsoft Access database that included the information located in the paper finding aid and then linked each event to the text location in The Public Papers of Lyndon B. Johnson. The database included additional fields to provide information, such as the file name and length, about the digital transferred files.
Another task during the practicum was to research options for outsourcing the reformatting project. While considering the specific needs of the library, I devised a questionnaire to follow when contacting a company about its reformatting services. Some of the library's specific needs and issues included: formats available for end product, quality of end product (file size and CD quality), experience with references, storage conditions, security issues, costs, and any additional services offered. One important factor was that the library wanted the work done locally. I reported my findings to my supervisor Philip Scott, Audiovisual Archivist at the LBJ Library and Museum, who presented the information to the Library Director, Betty Sue Flowers. The decision was made to perform the digital transfers in-house and I was offered the opportunity to do the work.
With the decision to do the reformatting in-house, I was asked to develop standards and procedures for the library. It was very difficult to decide what standards to develop because very little guidance is available and no "official" audio reformatting standards exist. There were several important factors I considered when developing standards for this specific project, including:
Nature of material—spoken word
Type of material—full track polyester reel-to-reel audio tape
Condition and sound quality of the tapes—good
Size of the collection—approximately 416 hours of materials.
The final standards and procedures were based on the needs and concerns of the library. We decided how to set appropriate sound levels for the event and then developed procedures to perform this task consistently. Standards to be developed included file size and file naming structures. We had to determine both the bit depth and sampling rate for the file size. We also had to decide how many generations of digital files to make and if preservation backup copies should be made on CDs. I researched the standards and procedures used by other archival institutions working on audio reformatting projects and looked for projects that were reformatting similar materials and institutions with national recognition. Conservation Online's Preservation of Audio Materials Web page 1 was a great source for information about audio preservation. The primary institutions I consulted included:
The Library of Congress: Audio Visual Prototyping Briefing Document and Digital Conversion Statement of Work. 2
Michigan State University: Spoken Word Project. "Working Paper on Digitizing Audio for the Nation Gallery of the Spoken Work and the African Online Digital Library. " 3
The Colorado Digitization Program, Digital Audio Working Group: Digital Audio Best Practices Version 1.2, May 2003. 4
With any type of preservation, the goal is to preserve the material as closely as possible to the way it was originally created. Reformatting is a very invasive type of preservation. It results in material being transferred from one type of medium, in our case analog reel-to-reel audio tape, to another type of medium, digital wave files. This transfer inherently changes the material. The primary problem when determining what to do with audio materials is that currently an archival medium is not available for audio materials. However, for this project it was already determined to transfer the analog reel-to-reel tapes to digital files. A major benefit with digital files is that once the material is transferred, further sound quality will not be lost during additional digital transfers.
I had to develop a procedure for determining where to set the sound levels on the reel-to-reel machine and on the computer when performing the digital transfer. The procedure included the following three requirements: (1) the preservation copy should be a flat transfer of the original, (2) no digital changes would be made, and (3) the sound level settings for the transfer would be the only element adjusted. For each event the sound levels would be set so that the loudest part of the event was just below clipping, and levels would never be adjusted during the event. Sound levels were checked by listening to the event and setting the levels on the reel-to-reel machine so that the VU meter levels were just below peaking. Then on the computer, the Line-In volume of the sound card was adjusted to make sure that the meters in Sound Forge also did not peak.
The next major decision was to set standards for the number of audio file generations to create and their file sizes. Some of the main issues and questions considered in making this decision included:
How many generations are needed to insure the preservation of the audio materials? More generations cost more money because they demand more storage space on a server and more materials for CD backup copies.
Digital capabilities are always improving so what is available now will relatively soon be out of date. For preservation purposes, the field generally tends to fall on the "high" end of the file size choices.
Larger file sizes cost more money because they require more processing time and storage space.
Is 24-bit justified over 16-bit because it helps to overcome imperfection in the digitization process? 5 Our tapes are in good condition and have relatively good sound quality.
Is 48 kHz sampling rate justified over 44 kHz sampling rate because of the material type? For the same reason is 96 kHz not justified for audio tape? The highest frequency pitch that a digital audio sample can hold is one-half of the sampling rate. 6 Thus, since analog audio tape is unable to capture frequencies above 24 kHz, a 48 kHz sampling rate has captured every thing possibly recorded on the audio tape. This reasoning seems to make 48 kHz preferred over 44 kHz and 96 kHz unnecessary.
We also had to weigh the benefits of higher file formats for our material, the spoken word. We performed an audio listening test to determine if differences could be heard between the different file size formats. The results were unclear; overall, differences were not readily noticeable. We devised standards based on the library's specific requirements and capabilities as well as our research of other institutions' standards. While we understood the importance of having a flat copy for preservation purposes, we were also hoping to address the issue of access. We decided to produce a reference copy so that, when necessary, we could adjust the sound level of the speech to make it easier for the listener to hear.
We decided to make three generation of files to be kept on the server:
Preservation Master: The file taken from the direct transfer of the reel-to-reel tape.
Uncompressed WAVE file format
24-bit depth and 48 kHz (kilohertz) sampling rate
No enhancements
Reduced Master: Produced from the Preservation Master—in Sound Forge the file will be digitally reduced from 24 to 16 bit and the sampling rate will be reduced from 48 to 44 kilohertz.
Uncompressed WAVE file format
16-bit depth and 44 kHz sampling rate
No enhancements
Reference Service Duplication Master: Produced from the Reduced Master. While listening to the original transfer, any noticeable sound level problems will be noted and then digitally corrected using the volume process in Sound Forge.
Uncompressed WAVE file format
16-bit depth and 44 kHz sampling rate—same as reduced master
Volume adjusted when necessary to make the speech a uniform sound level
This generation will be used to make copies to sell to the public and eventually will be used to make MP3 files for web use.
In addition to the three generations of files to be kept on the server, the decision was made to make three CD copies to serve as preservation backup copies:
Preservation Master: Data CD containing 24-bit and 48 kHz WAVE files. The WAVE file is uncompressed with no enhancements; it will serve as backup to the Preservation Master file on the server.
Audio CD: Made from the Reduced Master server files, it will be produced in "red book standard," also known as "CD standard." If nothing else survives, it is thought that one would at least be able to find a CD player that would recognize this copy.
Reference Service Duplication Master: Data CD 16-bit and 44kHz WAVE files. This copy will have volume adjustments and will serve as backup to the Reference Service Duplication Master server file. The main reason for this backup copy is to preserve the work and time involved in performing the volume adjustment.
The last standard we wanted to develop was a file naming structure. While the exact file naming structure we developed is not necessarily important, it is crucial to set a standard for naming that is consistent over time. As with all other standards that are developed, naming structures need to be documented for future users. The specific formula used to develop the file name for the LBJ Library was as follows:
Example: WHCAAUDSP004t1pm.wav WHCA - White House Communication Agency, that created the material. AUD - Audio material type SP - Speech President collection 004 - tape number (3 numeric digits because tape number go into the hundreds) t1 - track number (collection contains tapes with 1 to 7 tracks per tape) pm/rm/rsd - file generation (preservation master, reduced master, reference service duplication master) .wav - file format
For the project, we used Gold Mitsui CD-Rs with a white inkjet printable surface for backup copies. The file name was printed on the CDs with black ink. The digital files are being stored on a server with access limited by password. The workstation set-up was put together by Fletcher Burton, Head of Technical Services. The specifics of the workstation are as follows:
Reel-to-Reel Machine - Otari MX5050 Pre amp - Peavey 10/4 External sound card - Creative SB Extigy - Sound Blaster Dell Computer - Windows XP operating system, 1.4 MH processor and 1 gig HD CD printer - dedicated to only printing the CD surface (maker unknown) Jewel Case printer - HP 6122 CD Burner - Que Fire Transfer software - Sound Forge 5.0 Burning software - Nero - Burning Rom Printing software - Sure Thing CD Labeler used for both the jewel cases and the CDs.
Once there were general standards to follow and the equipment and software was set up, I began developing the actual transfer procedures. I started with a rough outline of each process in the procedure including (1) making the transfer, (2) burning the CD, and (3) making the jewel cases. Then I began to develop detailed step-by-step instructions on how to complete each process. The procedures were developed through a lot of trial and error. As I gained more familiarity with the equipment, the computer programs, and the collection, it became easier to determine the workflow. Three archival cabinets were purchased from Russ Bassett to store the three copies of the CDs that were being produced by the project. Two of the cabinets are located in different areas on the 3rd floor and the third cabinet is in the AV research reading room on the service level, one level below the first floor. When I left the project, I had completed 163 tapes including 454 individual speeches. One of my fellow colleagues, Sarah Cunningham, has taken over the project.
In the future, the library plans to sell the digital audio files that are being produced as part of this project. Currently reference requests for the Presidential Speeches are being performed in the technical services department where they transfer the audio file from reel-to-reel to digital only on a demand basis. In the future, digital files created by this project will be used to fill reference requests. A CD will be made from the RSD (reference service duplication) server file into the format chosen by the patron. We are considering offering the following file formats: MP3, CD Audio, and a Data CD of the WAVE file. All requests will be produced from the RSD file, which has sound level adjustments, unless specifically requested otherwise. Currently, we are not making MP3 copies of the audio file because we are unable to have the sound files up on the Web; however, the library plans to make these sound files available. For the time being, MP3 files will be made on an as needed basis. The library also plans to make the digital finding aid available online. Staff would like to have space on the Web server designated for some of the audio files so that they will be accessible and downloadable online. Since this collection of speeches is in the public domain, the library does not have to worry about copyright issues. The Audio-Visual (AV) Archives division would like to purchase an AV Archives server that would act as a preservation server for all audio and photo files created by the division. This would allow AV Archives to gain better control of their preservation digital files. Currently the audio files have been allotted a certain amount of space, with password protection, on one of the storage servers for the library. The server is located inside the library's firewall. The library currently has several servers; one is outside of the firewall for Web-related information and the rest are inside the firewall and are used for daily activities and storage. Once this project is completed, there are many other audio collections within the library that still need to be transferred, including Lady Bird's speeches and LBJ's vice-presidential speeches. The LBJ audio reformatting project has provided a unique learning experience and hopefully will provide useful information for other institutions looking to reformat their audio collections.
Conservation Online. Audio Preservation http://palimpsest.stanford.edu/bytopic/audio/.
Library of Congress. Digital Audio-Visual Preservation Prototyping Project. 23 March 2001. http://lcweb.loc.gov/rr/mopic/avprot/avprhome.html#coll.
Michigan State University. Working Paper on Digitizing Audio for the Nation Gallery of the Spoken Work and the African Online Digital Library. http://africandl.org/bestprac/audio/audio.html.
Colorado Digitization Program, Digital Audio Working Group. Digital Audio Best Practices Version 1.2. May 2003. http://www.cdpheritage.org/resource/audio/documents/CDPDABP_1-2.pdf.
Ibid
Ibid.
© 2004 Jill Hawkins