Home > Technical Writings > Digital Talking Books: Progress to Date > Work Accomplished to Date

NLS Technical Writings

Digital Talking Books: Progress to Date

Work Accomplished to Date

May 2002

previous --- next

NLS has been working on the development of a digital talking book (DTB) system on many fronts since the 1998 publication of Digital Talking Books: Planning for the Future. The following section describes NLS's progress in those efforts, which are grouped here into three categories: Building Standards and Creating Management Tools, Creating a Digital Collection, and Designing a Player. Each project discussed here includes a reference to the step(s) under which it falls in the Twenty Step plan laid out earlier in this document.

1. Building Standards and Creating Management Tools

ANSI/NISO Standard for Digital Talking Books (Step 1 in "Twenty Steps")

In December 2001, members of the National Information Standards Organization (NISO) voted to approve "Specifications for the Digital Talking Book" as an American National Standard. On March 6, 2002, the standard was approved by the American National Standards Institute (ANSI) as ANSI/NISO Z39.86-2002.

The standard is the result of a nearly five-year effort by an international committee representing a broad range of stakeholders working to provide alternative-format materials to print-disabled readers. NLS led the effort begun in 1997 all the way to final adoption of the standard. It chaired the NISO Digital Talking Book Committee, organized and ran the meetings, created and managed work groups, wrote much of the standard from committee input, and edited the entire 125-page document.

In this complex effort NLS's partners were the Association of Specialized and Cooperative Library Agencies (of the American Library Association); American Council of the Blind; American Foundation for the Blind; American Printing House for the Blind; Blinded Veterans Association; Canadian National Institute for the Blind; the DAISY Consortium; Hadley School for the Blind; Assistive Devices Industry Office-Industry Canada; IsSound Corporation; National Federation of the Blind; Recording for the Blind and Dyslexic; Talking Book Publishers, Inc.; Telex Communications, Inc.; VisuAide, Inc.; and the World Blind Union.

The ANSI/NISO standard is built on specifications and requirements formulated by blind and visually impaired users, who were heavily involved in every aspect of the development. During the process, the full committee met eleven times across the United States and Canada, and smaller working groups met countless times both in person and in conference calls and e-mail. As far as possible, the standard was built on existing standards and specifications. Thus programming skills and software tools developed for other purposes can be applied to the DTB world.

The standard is important because it makes possible a powerful, flexible reading system that adapts easily to different types of documents and different users. It provides a framework under which a person or agency can produce DTBs ranging from a very simple novel to a long, complex reference work or textbook. And it allows users great flexibility in how they read those DTBs-some will want a straight linear reading experience, while others will need sophisticated functions, among them random access to sections of the DTB, the flexibility to turn on or off selected elements (e.g., footnotes), and the capability to set bookmarks, highlight portions of text, or do keyword searches. So this single standard addresses the needs of a variety of agencies serving users with a wide range of reading needs.

A DTB is a collection of electronic files arranged to present information to the reader via alternative media. The most common medium, of course, will be human speech. However, a DTB produced in accordance with the ANSI/NISO standard can include a file that presents the contents of the document in text form, thereby permitting output via synthetic speech, refreshable braille, or visual display (e.g., large print). Because of the large expense involved in creating text files, NLS is not likely to produce many DTBs incorporating them. However, the capability is there when needed.

An aspect of the standards effort worthy of mention is the NISO/DTB committee's collaboration with the DAISY (Digital Audio-based Information System) Consortium, an international body established to develop specifications and tools for the production and delivery of DTBs. When the NISO committee began working, there was fear that it was trying to displace the DAISY group. DAISY was invited to provide representatives to the NISO committee and did so, greatly expanding the committee's international reach. It became evident very soon that sharing resources would benefit both groups.

It was agreed that the NISO committee would focus on an XML-based standard that would provide capabilities not possible in DAISY's HTML-based specifications. (XML is becoming the successor to HTML as the language of the World Wide Web.) The expertise DAISY developed in creating the first DTB specifications was invaluable to the NISO work. As stated in the acknowledgments section of the standard: "It is no exaggeration to state that without their groundbreaking efforts and their ongoing contributions to Committee work, this standard would not exist in anything like its current level of sophistication."

Life-Cycle Cost-Analysis Model (Step 3)

In September 2000, NLS took delivery of a computer-based cost-analysis system that allows comparison of costs for the current audio cassette program with costs likely in the deployment of a new technology. Produced by the Northrop Grumman Corporation under contract to NLS, the Life-Cycle Cost-Analysis Model is being used by NLS to compute system costs associated with a variety of DTB distribution scenarios.

The model looks at all costs expected to be incurred in the production, distribution, and maintenance of audio books, magazines, and playback machines by NLS and its multistage centers, and the network of cooperating libraries. It computes comparative costs for a ten-year period of both the current program and any alternative scenario.

The model is a dynamic application running within a spreadsheet program. It was developed by an interdisciplinary team of NLS personnel thoroughly knowledgeable of the current audio-book system. NLS engineers, librarians, managers, and production analysts met with a contractor team that included cost analysts and software engineers. The joint team identified 101 distinct data elements, along with their interrelationships. For NLS costs, exact expenditures are captured for the "current year," which rolls forward every twelve months. For the network, the basis was formed by the exhaustive study of activity costs conducted in 1989-1990. Network costs were brought forward using standard inflation factors published by the U.S. Commerce Department.

The model provides a way to model alternatives rapidly. Costs can be expressed in terms of dollars per reader per year, dollars per circulation, etc., or they can be reported as year-by-year totals over a ten-year period, reflecting the effects of anticipated inflation. The model will rapidly recompute the effects of say, halving the cost of the book media or doubling the reliability of the playback machine.

The model, because it uses many estimated values, is not expected to produce precise values for overall system costs. Rather, the model will be used not only to develop comparisons between alternatives but also to determine which factors have the greatest effect on overall costs. It will also ensure that NLS and the network examine all the elements that contribute to the total cost of a system and that none are overlooked.

As part of their contract, Northrop Grumman compared actual costs of the current analog cassette system with projected costs of a CD-ROM-based system. This exercise, which included NLS staff in the gathering and analysis of data, demonstrated how the tool would be used and gave NLS insights into its power and flexibility. The scenario indicated, however, that the CD-ROM alternative would be considerably more expensive than the cassette system, largely because of the higher costs to purchase and maintain CD-ROM players.

Digital Long-Term Planning Group (Step 8)

During 2001, NLS established a group made up of consumer representatives and network librarians, who, along with appropriate NLS staff, will contribute to planning for the deployment of digital information technology throughout the network in years to come. In order to guarantee continuity and responsiveness to ever-changing prospects, the group will likely meet once or twice each year through the next five years. It held its first meeting at NLS October 24-26, 2001. While the group's mandate covers a range of subjects related to digital technology, at least half of its time will be devoted to planning the implementation of a digital talking-book system.

At the first meeting, NLS presented detailed background on current NLS DTB initiatives. The presentation described progress in the development of the DTB standard, production of new digital recordings as well as selection of analog titles to be converted to digital format, development of a PC-based DTB player, identification of critical design issues for a DTB system, and analysis of digital distribution options (such as CD-ROM, Internet, or solid-state memory). The group then discussed in detail three possible scenarios for distributing DTBs on solid-state memory cartridges.

In the discussion of distribution scenarios, NLS introduced the Life-Cycle Cost-Analysis Model to the committee, reviewing some of the previously mentioned 101 elements and demonstrating how the model can be used. The group was established partly to involve members in developing data values for network costs to be used in the cost-analysis model, ensuring that NLS develop figures that are as accurate as possible. Another key reason for bringing the group together was illustrated during distribution discussions, as network and consumer representatives brought forth issues critical to their respective communities. This group will play an important role in the design of the digital talking-book system, ensuring, in concert with NLS, that it is cost-effective and also meets the needs of libraries and consumers alike.

2. Creating a Digital Collection

Digital Recording (Steps 6A and 7A)

The NLS recording studio has been experimenting with a variety of digital recording systems for several years, an effort serving a number of purposes. First, it has enabled NLS staff to maintain an understanding of the kinds of available software and hardware as well as the trade-offs associated with each. Second, NLS studio staff gained experience with features that are entirely different in the digital domain from in the analog world-notably working with PCs instead of the standard five-button interface on open-reel recording decks, and learning how to manage the very large audio files demanded by a DTB. Third, the effort has offered to the recording studio the chance to begin developing a collection of digital recordings. And fourth, it has provided vital first-hand information to NLS staff and contractors who are designing a low-complexity digital recording system (described below) for network libraries.

Two of the three studios in the NLS recording complex have been outfitted with digital recording equipment. The staff, many of whom were not experienced PC users, have learned to work efficiently in the digital environment. At this point, the studio is producing only digital master recordings; it is not creating fully-functional DTBs. Simple and efficient-to-use software capable of producing ANSI/NISO-compliant DTBs will be installed once it is available, and the staff trained on the new aspects of DTB production.

Beginning with The Complete Fairy Tales of Charles Perrault, the studio has digitally recorded over 50 titles to date. Some of these have been added to the current cassette collection, while some have also been developed into DTBs using a combination of experimental automated tools and manual efforts. These test DTBs, selected because they include a wide variety of structures and such text elements as footnotes, indexes, glossaries, etc., have been used with the PC-based player (described below) in order to test the player's functions as well as to demonstrate the feasibility of different features.

NLS audio-book contractors began recording a portion of their assigned titles digitally in fiscal year 2002. NLS contracts stipulate that at least 10 percent of the titles produced in 2002 be recorded digitally. In 2003, contracts will require half of all titles recorded to be in digital form; in the following year, all new titles will be recorded digitally. For the next several years at least, all titles-whether recorded in digital or analog format-will continue to be distributed on analog cassette. However, the digital recordings will be available for distribution when NLS begins to circulate books on a digital medium.

Low-Complexity Mastering System (Step 6B)

NLS has developed specifications for a recording system especially designed for network libraries' use. After reviewing a broad range of digital recording software products and gaining experience with several of them in our studios as described above, NLS staff concluded that no existing digital audio recording system meets NLS requirements for efficiency and quality.

The majority of such digital audio systems are designed for recording music, are highly complex, and assume that following a recording session considerable time will be spent editing the sound track(s) to produce a finished product. In contrast, a recording system designed for library use must be straightforward and easy to use, making efficient use of the narrator's and monitor's time. NLS staff's specifications for the Low-Complexity Mastering System are based on analysis of existing commercial and talking-book-specific systems, discussions with both contractors and network library staff, and focus-group sessions with NLS studio staff. Development of the system is underway and a prototype has been installed in the NLS studio.

Experimental DTB Production Tools (Steps 6B and 6D)

As mentioned above, NLS will only produce a limited number of DTBs that include a full text file. However, the presence of the text file will greatly increase the usefulness of some titles, making cookbooks or reference books, for example, easier to read and richer sources of information. Producing full-text/full-audio DTBs can be a complex process. It requires the production of a very detailed synchronization file linking the audio file to the text file. In one approach to creating this file, the monitor must perform frequent, repetitive tasks during the recording process. To free the monitor of this burden, NLS is experimenting with software that automatically captures the necessary timing information (the times at which each word in the text file begins and ends) and then generates the synchronization file. NLS works with several different alignment tools (software that captures word timing) and has developed a program that automatically generates the synchronization file-an approach that is still experimental but shows great promise for simplifying the production of full-text/full-audio DTBs.

Converting Analog Titles to Digital (Step 5A)

The current analog cassette collection contains over 50,000 titles. Some portion of this collection will need to be converted to digital format in order to be available to readers when audio books begin to be distributed on a digital medium. What proportion of the collection should be converted? To answer that question, the NLS Collection Development Section (CDS) staff has begun a multiyear analysis of the cassette collection, identifying the titles appropriate for conversion.

In 2001, CDS chose an initial 1,000 cassette titles to be transferred from analog to digital format. Selection of these titles created a cross-section of the cassette collection, with genres selected in proportion to their representation in the full range of cassette titles. For this first group, CDS selected titles it judged to be of most enduring value. The process will be repeated in 2002 and subsequent years, with the annual number of selected titles increasing as needed to meet production goals. This approach assures that whenever NLS begins to convert titles to digital format, a broad, representative range of titles will be available to readers.

3. Designing a Player

PC-Based DTB Player (Step 2)

NLS has designed and programmed a software-based digital talking-book player that runs on a personal computer.

This software serves several purposes. First, it enables NLS to test whether concepts developed during the standards process actually prove to be useful. For example, do navigation techniques really work as planned?

Second, it helps NLS to determine whether production tools are creating DTBs that comply with the standard. (Conversely, the sample DTBs help identify bugs in the software-based player.)

Third, and most important, the PC-based player allows NLS to experiment with different user interfaces; that is, the control and command arrangement enabling the user to read a digital talking book. The software is written in a manner allowing quick and efficient updates in response to user suggestions.

It is likely that NLS will eventually develop two kinds of DTB player-a basic player designed for linear reading and a more sophisticated device aimed at users wishing to randomly access or efficiently navigate complex documents. At this writing, the software player includes most of the functions expected in the basic player, among them play/pause; jump forward or back by paragraph, section, or chapter; set and go to bookmarks; and follow links (e.g., jump directly from the table of contents to the fifth article in a magazine). Several additional features planned for the advanced player are also present in the player, such as the ability to highlight segments of text and "turn off" selected types of material (such as footnotes), so that the player automatically skips over them. The basic functions of the software-based player were programmed by NLS staff and later enhanced by a software firm under contract to NLS.

Player Design Contest (Step 9)

NLS has teamed up with the Industrial Designers Society of America (IDSA) to sponsor a contest for industrial design students, challenging them to design the shell of a digital talking-book player. Students at any of the 55 IDSA-accredited industrial design schools were eligible to participate in the January 1-May 15, 2002, contest. NLS provided a list of features desired in a DTB player, drawn from the initial work on the DTB standard. The list offered students a range of often conflicting constraints which they were obliged to attempt to satisfy.

Students were asked to design the exterior of the device, including the layout of controls, configuration of speaker and jacks, and the design of tactile and visual labels. NLS probably will not use any single design, but will most likely draw interesting ideas or design features from multiple entries. A jury assembled by NLS and IDSA will meet in June 2002 to select contest winners.

Michael M. Moodie
Research and Development Officer

Prologue --- Twenty Steps to Next-Generation NLS Technology --- Work Accomplished to Date --- Nine Tasks to Implement the Use of Digital Talking Books --- Digital Braille: Web-Braille Puts Braille Books on the Internet --- Bibliography

Library of Congress Home      NLS Home     Comments about NLS to [email protected]     About this site      Legal     Comments about this site to the NLS Reference Section

Posted on 2013-06-28