Back to Artificial Language Lab Web Site
Speech Facilities for the Reading Disabled



Using simple text-to-speech output and alternative mechanisms to tackle mountains of text.



Carl Friedlander


Brain cancer and associated brain surgery can result in a variety of physical, psychological, and functional disabilities. To some extent, these disabilities are manageable using soft ware or hardware support systems. In my case, I have been using aids to help increase my reading speed. Like a sight-impaired reader, I need help reading text. However, unlike a sight-impaired reader, I can easily see objects, images and text components and I can easily operate my keyboard and mouse. However, the length of time it takes me to verbally name an object or silently read text is haltingly long.

My specific functionality loss includes a general reduction in reading rate with a severely reduced rate for reading "out loud." I easily comprehend email and research articles. However, in order to read information in a text-only format that is not obtainable where it can easily be produced in machine-related, sound-producible form, I find myself faced with information much more difficult to access. As a result, I am constantly looking for ways to better utilize the computers I use that produce spoken versions of text. I do not claim to have examined all the available machine-readable systems. But I can tell you about what I use, what has worked, and what enables me to function well.

Fortunately, a significant amount of my day-to day activity and the related information I need on a regular basis is directly available in machine-read able form. This information includes email messages, presentation materials, various program reports, PowerPoint image files, and spreadsheets. Multimedia presentation of objects as both text and sound provides rapid recognition of text-based objects that would otherwise be difficult to parse or read. I am also fortunate that a significant amount of my training was with John Eulenberg and Mort Rahimi of the Artificial Language Laboratory in East Lansing, Michigan. Much of our research work was directed toward the development of speech producing prosthetic devices to aid individuals with neurological, sensory or motor restrictions. My connection with that work was discontinued approximately 12 years ago, when I started working on other computer-related problems, including parallel processing and operating systems. I find myself dredging up the work that I performed long ago.

Problem Solving

My current day-to-day activities primarily require me to work at a computer system. I produce reports, utilize spreadsheets to determine the status of various programs, occasionally write functional code to evaluate the feasibility of an idea, and often develop slides for use as a presentation. I am working on the development of papers and system description or documentation. In addition, I constantly read material provided in either text or document form. I read new research papers, new program activity descriptions, and project descriptions. My recent brain surgery has encouraged me to develop alternative mechanisms for dealing with the mountains of text that come my way.

In order to perform more effectively, I decided I needed to utilize one of several ways to get my documents read to me. Fortunately, a number of individuals have been working on the problem of making computer systems more accessible to the blind. One of the most familiar voice-producing systems is the Kurzweil Reader. Though originally produced several years ago, the system is still available. The text-to-speech research work that was part of the original speaking machines lives on. Existing systems are available as a collection of speech-producing software programs. Developers who have attacked this problem have done so by building text-to-speech-producing agents for execution as part of computer systems.

M y first plan of attack was the integration of existing text-to-speech processors as part of the existing systems. The Macintosh I use provides simple text-to-speech output. A large number of text-to-speech systems exist, including WRITE:OUTLOUD. by Don John ston, and OutSPOKEN, by Berkeley Systems, and produce reasonable quality text-to-speech output. These systems are usable for reading the output of online documents. Virtually any document can be passed through one of the simple Macintosh applications. The existence of this text-to-speech facility on the Macintosh allows me to read much of the data available.

Day-to-Day Activities

The Macintosh SimpleText program is a software product provided as a part of all Apple Macintosh operating systems and applications. It provides text-to-speech output for any text elements displayed within its input boundaries. Transferring text to SimpleText requires users to copy the text they wish converted, then paste that text into the text space of the currently available SimpleText area. The "speak" command translates the pasted text to a spoken version of the text. I have found this facility extremely useful for reading my email. Even better would be a version of my email facility that directly provided text-to speech capability. I have not yet managed to completely solve that problem. However, there are programs (such as outSPOKEN) that come close to providing that facility. Systems developed and intended to support visually impaired users provide effective text-to-speech translation. I currently use a number of these facilities in order to improve my performance and the utility of my system .

O ne of the biggest problems remains that of reading long paper documents or articles. However self-sufficient I would like to be, sometimes it is necessary to have others read to me. However, I am trying to be self sufficient. Although it is true that journals are increasingly distributed or advertised by Web page, they are not fully distributed in machine-readable form. For example, I recently received a monthly copy of a journal. Though I could read portions of the magazine by pulling up copies of the text where it was part of a Web page, there were large portions of the magazine that could only be read by having some one else read them to me. Ideally, I would like to be able to receive a machine-readable version of the magazine in addition to the paper version. Then, I could select portions of the magazine to be read to me as needed.

Conclusion

Prior to becoming a more capable user of the text generating facilities of the Macintosh and the many other speech-producing facilities of the Macintosh, I was dependent upon others for aid in obtaining more rapid reading and information gathering. Using speech-producing facilities has enabled me to collect information and utilize it in a more rapid manner. I now look forward to accessing more of the contents of materials. I look forward to enjoying reading as I have in the past. Critical to that goal is the development of a mechanism for reading scanned text. <c>

Acknowledgments

The author wishes to give special thanks to his wife for her help in both preparing this article and in providing improved access to a number of documents. Additional thanks go to John Eulenberg who provided pointers toward the Mac applications that have become critical facilities of my daily use.

CARL FRIEDLANDER (friedla@isx.com) is vice president and chief scientist at the ISX Corporation, Westlake Village, CA.


ACM©0002-0782/97/0300 $3.50

COMMUNICATIONS OF THE ACM March 1997/Vol. 40, No 3