Braille Production the DAISY Way

by George Kerscher, Research Fellow
Project Manager to the DAISY Consortium
1203 Pineview Dr.
Missoula, MT 59802
USA
Email: kerscher@montana.com
Phone: (01)406/549-4687

ABSTRACT

The DAISY Consortium is developing the standards for the next generation of information technology for persons who are blind and print disabled. (http://www.daisy.org) While DAISY is famous for development of the Digital Audio-based Information System many people do not yet know that the specifications are comprehensive and include Etext and Braille in the production techniques. This paper will explain how high quality Braille can be produced using data prepared in the DAISY / NISO DTBook 3.0 XML specification currently under development.

ACKNOWLEDGEMENTS

I would like to thank Joe Sullivan from Duxbury and Aaron Leventhal from the Braille Planet for their contributions. Joe and I have worked together for years promoting descriptive markup in files used in production of Braille. Joe and I have worked on many committees together developing this philosophy. Aaron graciously offered to review and assist in the development of this paper. Much of the content of this paper is based on the long relationship I have had with these two organizations and the fine work they have done.

Three types of DAISY Digital Talking Books (DTB)

The DAISY specification describes three types of DTB. One type is the full text and full audio (hybrid), another is a table of contents with full audio, and the final version is full text with no human recorded audio (synthetic speech would provide the audio component). Two of these three types of books are of interest to the Braille producer. In the full text with audio, and the text only versions, the marked up text essential to Braille production is available. Braille producers can use this data as the starting point for their Braille production process.

NOTE: DAISY Consortium representatives are involved in the National Information Standards Organization (NISO) work on Digital Talking Books (DTB). Some organizations not part of DAISY also participate in these developments. This NISO and DAISY joint effort is producing the formal Document Type Definition (DTB) which in this paper is being called the DAISY / NISO 3.0 XML DTD. Please don't be confused by the terms DTB for digital talking book and DTD which stands for document type definition!

In the DAISY DTB there is a clear separation of the audio files and the text files encoded with the eXtensible Markup Language (XML). The linkage between the two separate files is the Synchronized Multimedia Integration Language (SMIL) files. This means that a Braille producer can simply copy the XML files and move them directly into the Braille production process.

What does the XML file contain?

The text of the whole book should be marked up in accordance to the DAISY / NISO XML 3.0 DTD specification. This means that each heading, paragraph, list item -- essentially all the elements that make up the book are clearly and explicitly marked. There is no mistaking what is a heading or paragraph, or page numbers, because there are unique tags (marks) that identify the elements. The actual text is present and at the beginning of the portion of text is a tag and a matching end tag at the end of the text. There is no guesswork at all; everything is clearly identified.

The markup is similar (but much better) to what we see on the World Wide Web in HTML. Current HTML web sites use a lot of constructs designed to force appearance while failing to convey, or even obscuring, the true document structure. The visual constructs that are problematic in HTML are not allowed in the DAISY /NISO 3.0 XML DTD. However, the basic concepts are the same. For example in HTML and in our XML DTD a <h1> is found at the beginning of a heading. At the end of the heading is a </h1>. This clearly identifies the text between the begin tag and the end tag as a heading. The constructs from HTML that convey proper document structure are preserved in our DTD, but purely visual presentation are moved outside the DTD to styles where they belong.

How Do XML Files Fit Into the Braille Production Process?

The Braille producers start by taking the completed XML files. These files should already have descriptions of figures and other graphics. These files are then "IMPORTED" into their Braille translation software such as Duxbury or MegaDots. The Braille translation software understands the markup in the files and translates the tags into the codes used in their systems. At this point the files are ready for examination by an experienced computerized Braille transcriber. The work that remains is to look for difficult transcription problems.

Importing of the DAISY files is much different from other import facilities provided within Braille transcription software. If you do not use a DAISY / NISO DTB XML file as input, the translation software must "GUESS" at the various elements. For example, if the translation software sees something on a line by itself, it guesses it is a heading and formats it accordingly. There are hundreds or perhaps thousands of guesses the software must make. Using the DAISY / NISO DTB XML files as input, eliminates the guesswork involved. This is a very straight forward translation process. I do not mean that this is simple software for the Braille translation companies to write, but when it is completed it is extremely accurate. One should not compare the importing of other types of files with DAISY / NISO DTB XML files!

The tagging provides the information for formatting of the Braille files. Knowing the files were previously marked up, spell checked, figure descriptions added, and proof read takes this burden from the Braille producer. This leaves only the most difficult of problems for the computerized Braille transcriber to address. This maximizes the Braille transcribers time and allows them to produce more Braille with fewer resources.

It is expected that the most difficult problems will still need the attention of the Braille transcriber to produce perfect Braille. In many cases the Braille output will be acceptable for many situations, but for textbooks used in school especially at early grades the Braille transcriber's attention will be necessary. Tables will be a problem that the transcriber will need to address. Also special rules of Braille for the front matter will probably need to be manually formatted. There are also some items that require human intelligence to interpret. For example the letter "A" may mean the letter, or it may refer to "Group A". The Braille transcriber will need to treat these differently, but the computerized Braille translation software will greatly assist them in this process

What Braille Translation Software Supports the DAISY /NISO XML 3.0 DTD?

The DAISY / NISO DTBook3 DTD is under development and testing at the time of this paper submission. We have several Braille software production companies reviewing the file specification. With the spread of DAISY throughout the world, we expect that most Braille translation software will support the DAISY / NISO XML specification or risk loosing large portions of their market share.

This Sounds Too Good to be True! How can this be?

There is no magic in this process. There is a shift in the production effort. Instead of spending countless hours on Braille data input and proofing, the time is spent on the XML data creation and proofing processes. What we see is the convergence of the production techniques to produce Digital Talking Books and the Braille production process.

How To Produce More Books in XML?

It is not easy to produce books in XML. We may find that we have just as much difficulty producing the XML as we have producing Braille today. The first benefit is that the DTB is actually better with full text and that the skills of the trained transcriber are used more efficiently. However, there is the prospect that publishers themselves will be able to provide files in XML!

Publishers still feel their primary product is the printed book. However, they see the trends moving to electronic books. Openebook (http://www.openebook.org) is a recent initiative among publishers, software developers and hardware manufactures to define standards for electronic books. Openebook efforts are underway to define XML notation that can be used to deliver electronic books on hand held devices such as the Rocket Book and Soft Book. These devices are in their infancy, but promise to become more popular in the future. These products have no market unless there are books that conform to the specifications. The DAISY Consortium is working to ensure a convergence of this mainstream technology and the developments within the DAISY Consortium.

Conclusions

1. DAISY DTB produced with full text is the perfect input for computerized Braille translation software.
2. The DAISY / NISO 3.0 XML files have already had figure descriptions added.
3. The DAISY / NISO 3.0 XML files have already been spell checked and proof read.
4. Modern Braille translation software will support sophisticated translation processes from DAISY / NISO XML files.
5. Many libraries around the world will elect to integrate their Braille production processes with their DTB production.
6. Publishers of traditional printed books will begin to move to a production process that supports XML.
7. If your library is not yet part of the DAISY Consortium, it is clear that it is time to join!