RFB&D'S NEW DIGITAL AUDIO SYSTEM: TAKING ADVANTAGE OF DIGITAL TECHNOLOGY
John Churchill
Vice President Operations
Recording for the Blind & Dyslexic
20 Roszel Road
Princeton, NJ 08540
Voice: (609)520-8067
FAX: (609) 520-7996
Internet: jchurchill@rfbd.org
Robert Rao
Marketing Manager
Recording for the Blind & Dyslexic
20 Roszel Road
Princeton, NJ 08540
Voice: (609) 520-8041
FAX: (609) 520-7990
Internet: rrao@rfbd.org
Web Posted on: November 22, 1997
Members of RFB&D's service have told us they prefer listening to human readers, but that it's difficult to find information on a audio cassette. Our members have also told us they enjoy the navigation and searching capabilities of our E-Text products, but listening to synthetic speech is fatiguing and less pleasurable than human speech. The concept of merging electronic text and digital human speech marries the strengths of these two technologies. A structured text file coordinated and synchronized with the digitally stored human recording will be the next generation of accessible materials from RFB&D. By indexing an electronic text file to its companion digital audio file, we leverage the navigational and searching attributes of E-Text and the pleasure of listening to a human reader from the digitally stored recording.
HOW BOOKS ARE DIGITALLY RECORDED
While it is possible to digitize existing analog recordings, the next generation recording device used by our volunteers will be a multi-media computer. The reader will speak into a microphone connected to a sound card in the computer and software will control the storage of the speech on the hard disk. The digital recording software also stores a continuous time indicator. These time stamps make it possible to know where each page begins or where a new section starts. The person monitoring the reader, or the readers themselves, will control the multi-media computer and simply press a key whenever a new section heading or page boundary is crossed. The time stamps of these locations will be stored for later use. These digital recordings can be of any length of time, depending upon the size of material read. RFB&D's books range from 1 to 70 hours in length. The reader and monitor will be able perform corrections as they do today. Digitally recorded sessions can be joined together, and the time stamps will be automatically adjusted. When the book is completed, there will be one large digital recording with associated time stamps that represents the beginning of each chapter, section, sub-section, and page break. These time stamps will be used to coordinate with the text file.
HOW THE STRUCTURED TEXT FILES WORK
The structured text, stored as HTML, represents all the information in the recorded book in a standard form. The actual text is stored as keyboard characters commonly known as ASCII. The chapter headings, sections, and sub-sections are stored in the file as HTML tags. There are also some HTML extensions that have been proposed for special purposes. For example the beginning of the ink print page 95 would have the mark "<ipp no=95>." Marking up the ASCII text with HTML tags allows for exact identification of locations in the ASCII file.
COORDINATING THE RECORDING WITH THE TEXT
Part of the manufacturing process will be to associate each heading tag and page tag with a time stamp in the digital recording. Our volunteers will look at the heading and listen to the recording. Time stamps can be adjusted, or new ones put in if they were missed in the original recording. This phase is a process of verification and fine tuning of the digital recording's time stamps. Most importantly, is that this process must be easy to use.
STORAGE
The digital audio recordings require a significant amount of storage space. While in the manufacturing process, files will be stored on high capacity computer hard disk drives. Once the book is completed and has passed our quality control, it will be "burned" onto a CD. Eventually, we may employ "Digital Versatile Disk" (DVD) technology. Use of this high capacity storage technology will allow us to store not only the digitized voice, but the text and coordinating information. This will be the archive copy.
We expect that from this archive, a more traditional CD-ROM version will be created that uses the latest in MPEG compression. It remains to be seen exactly how much we can store on a standard CD-ROM using this compression, but we expect to have 50 hours of very good quality sound. This compressed CD-ROM will be what is shipped to the RFB&D member.
USING A MULTI-MEDIA COMPUTER TO READ
For students and professionals, using a standard multi-media computer will be the primary method for using the digital text and audio book. Using navigational software currently under development, members will access our books in much the same way as a printed book -- through its table of contents. Using the arrow keys one would navigate to any section or sub-section. From there a simple keystroke launches the playback of the audio recording through their sound card.
In the above example, the structured text is simply used to navigate to the desired section of the digitally recorded book. More complete versions may include full text, which corresponds with the digital recording. At every synchronization point -- an HTML tag and an associated time stamp -- the visual presentation of the text is readjusted on the computer screen. In addition to the navigation features offered through the use of the table of contents, detailed searching can be performed on the text itself. To locate a specific string of text, the user simply types in the desired text and its position is located within the book. With a single keystroke, the synchronized audio recording is played. For those individuals who benefit from a combination of visual and audio input, this full text format will offer the ability to highlight individual words while they are being read through a speech synthesizer.
Although, RFB&D does not have immediate plans to develop a hand held digital book player, we may proceed in this area should our members' needs require one.
DISTRIBUTION
It will be necessary to continue to distribute our materials on cassette to support traditional tape players. Therefore, although we will begin producing digital masters, distribution of our materials will be available either on cassette, or CD. We will continue to support our traditional cassette users. However, migrating our library from analog to a digital/HTML based structure will position RFB&D to provide service through a variety of distribution channels, such as the Internet. Ultimately, on-line delivery of our educational materials tailored to our members' needs is our objective. We envision our members browsing our on- line catalog, selecting entire books, or excerpts and downloading their selections.
BOOK CONSTRUCTION FLEXIBILITY
Structured full text and full digital recording on one CD is highly desirable for students and professionals. However, production of this type of book will require more effort than a product that is created in either a digital or electronic format.
We expect that there will be variety of versions of digital audio books produced. Some books will have all features, that is full text with audio. Others will have only the table of contents and index in textual form with the full digital audio recording. Reference books, such as dictionaries will be prime candidates for straight E-Text production -- with no human voice recorded. This gives RFB&D the flexibility to meet the needs of our members with the features that are appropriate for the type of book.
TIMETABLE
RFB&D is proceeding with the development of the digital recording station and the playback software to be used with a multi-media computer. We expect to begin testing beta versions by Spring 1997. After several months of internal testing and evaluation, we intend to begin installing beta recording stations in our studios. RFB&D has 30 recording studios located throughout the country comprising over 180 recording workstations. We expect that conversion from analog to digital recording throughout our network will be a multi-year project. Plans for this change are currently underway.
OTHERS WORKING ON DIGITAL AUDIO
Throughout the world, other organizations are researching the use of this new technology. We hope that through a collaborative effort, international standards will be adopted to create a worldwide library of accessible material.