One of the greatest minds of the 20th Century is trapped inside a body that doesn’t work. Yet Steven Hawking has produced three books that change the way we look at our world. Without his VOCA, (Voice Output Communication Aid), Dr. Hawking’s powerful ideas would be locked away forever—unspoken, unwritten, and unknown.
In the same way, speech technology can change the way we view people with disabilities. Speech recognition software has enabled persons with writing disabilities and challenges to put their ideas onto paper. As we move into the new millennium, the world’s economies are leaving their resource base and rapidly changing to their new base of information and technology. Speech technologies are one way to ensure that a physical disability is no longer an obstacle to participation in our economy. However, speech recognition software isn’t an instant fix. It should be considered a part of the solution.
The following are stories of people with disabilities who use speech recognition software. All of these people needed special training and adaptations to use their software effectively.
Name: Ray Walker
Disability: Multiple Sclerosis (MS)
Computer: Desktop Pentium PC
Dictating Environment: Quiet home
Microphone: Andrea NC 80 duct taped to a goose neck lamp
Software: Dragon Dictate for Windows v3 and Office 97
Roadblocks: No use of hands, soft speech, short utterances, low vision and stamina
What made the difference: Good on site support for training and voice file maintenance, duct tape.
As a graduate student at Simon Fraser University, Ray Walker is completing his thesis on indirect communication in literature. When multiple sclerosis silenced his computer’s keyboard, he turned to speech recognition. MS made his speech soft, and communication comes in the small fragments that his lungs allow.
Ray and others like him need very special software and modified training to make it work. Some users with physical disabilities not only have lost the use of their hands, but often times speech is difficult for them to produce and others to understand.
For Ray, discrete speech input software is the only workable input option available. Ray needs complete access to the command and control features of his computer. Today this allows him to produce his written work independently.
When an environmental control system such as X-10 is added Ray will be able to use the telephone, turn on lights, fans, stereo and open doors using voice commands. Where there is a need is for complete command and control of the computer and a requirement to use environmental controls (such as X-10), Dragon Dictate 3 has no peer. Kurzweil Voice is a very good choice where command and control are of a lower priority or not needed.
On-site support is critical to Ray’s successful use of the product. His attendant can and does use Dragon Dictate. Equally important to being a computer expert, his attendant is a "Ray" expert. He knows when Ray needs a rest (even if Ray doesn’t). He also recognizes which words Ray can’t get past and need to be trained or entered manually. He tweaks the system when Ray is fatigued and the computer holds hostage the last few words that Ray needs. This teamwork is essential as Ray’s physiology can change on an hour-to-hour basis. Additional factors that affect training/performance include:
Mud on the Headlights
- Individual training of high frequency words
- Adjusting options such as word pause (word gap), quick talk, accuracy over speed
- Substitution of new commands and words for words and commands too difficult for user or not recognized. e.g. the command "whiteout" when "scratch that" is too difficult to produce.
- Compensation for respiratory and environmental noise Custom microphone configuration
- Seating, posture and attitude to facilitate speech production Therapeutic considerations for speech (breath control and verbal stamina)
- Adjusting font size and speech feedback for low vision users
- Maximizing hands free access
- Streamlining training and interface to minimum user needs
- Line amplifier (for speakers with low vocal intensity)
Rayis an award-winning writer. His short story, Mud on the Headlights, was the winning submission in a writing contest sponsored by the Canadian Broadcasting Corp. and performed over the air.
"When first diagnosed with MS, I felt very discouraged. I liked to drive and owned a snazzy new sports car. On a mid-winter nighttime trip through the dangerous Fraser Canyon, I was driving erratically. Obviously, the disease was taking over. My eyesight was failing and my driving days were at an end. Stopping –– with great reluctance and grief –– I gave up driving. My wife and I sat for a long time in silence while snow slowly covered the car. We were ‘parked’ in the ‘unknown’, emotionally adrift and indistinguishable from any other snowdrift... somewhere on the Trans-Canada Highway. Suddenly panicked by this realization, we frantically changed seats. As my wife ran round the front of the car,her thigh brushed the muck from the headlamp. Light shot down the road.I could see clearly! —Relentless, irrevocable tragedy was only "mud on the headlights."
Name: Laura Bridgeman
Computer: Laptop Pentium 166
Dictating Environment: At home and in seminar rooms at school
Software: Kurzweil Voice, Co:Writer and Kurzweil 3000
Roadblocks: Low literacy, little successful writing experience
What made the difference: Started with small controlled vocabulary with problem solving approach, intensive training and effective text reader. Desire to write. Ongoing support.
Laura Bridgeman is an award-winning athlete. At home, her wall and dresser display the ribbons and trophies she has earned. Laura works just as hard at her academics, but until recently there were no awards coming home from school.
Laura has dyslexia. Like others with dyslexia, she has great difficulty reading and writing the letters and words that make up our language. For Laura, the beginning of the 10th grade promised continued frustration and disappointment. The increasing requirement for reading and writing meant she would not be able to do her work independently. Something new was needed.
A trial using a combination of speech recognition software (Kurzweil Voice), word prediction with speech feedback (Co:Writer) and a powerful OCR text-to-speech program (Kurzweil 3000) have given Laura unprecedented access to reading and writing. This fall she received her first academic award for a paper and poster session she wrote on dyslexia.
For the individual who struggles with the written word, software that assists in reading and writing can open the doors to the world of literacy. It takes a combination of speech technologies and modified training strategies to achieve the best results. Technology-based solutions do not exist without motivation, ongoing support at home and at school, and an understanding of all the elements involved in the disability and the technology.
To unlock the door to independent writing, two types of software should be explored: speech recognition and word prediction. For speech recognition, it is recommended that the user start with discrete speech input software such as Kurzweil Voice or Dragon Dictate. These products provide the most manageable environment for users with low literacy levels and lend themselves to effective problem-solving when a literacy roadblock is encountered. The potential for quick training, high initial recognition levels and on-screen alternatives makes this a good place to begin. Modified training starts with a small, high-frequency vocabulary. The vocabulary items should focus on a user’s interest or a current curriculum area. This approach can demonstrate to the user his own abilities combined with the power of speech technology.
The biggest obstacle for speech recognition users with dyslexia is the inability to read and little experience with the writing process. For the user who, because of past failures or disability, hasn’t worked much with words, the best approach is often one word at a time in a controlled environment. This is best accomplished through discrete (one word at a time) products such as Kurzweil Voice and Dragon Dictate. Kurzweil Voice gets the nod for young users and Dragon Dictate for users with multiple disabilities.
In the simplest terms, the computer listens to you and displays what it thinks you said. The user must recognize if the computer guessed correctly. Since there are some environments where speech input is not suited (such as the classroom) Laura needs an alternative input method. Don Johnston’s Co:Writer (www.donjohnston.com) becomes Laura’s "pencil’ in the classroom. Co:Writer possesses a word prediction engine that provides speech feedback. The user begins to type the word she wants and based on spelling, grammar and word environment Co:Writer tries to predict and enter the word into the word processing document. Kurzweil 3000 "reads" her textbook and handouts when they are placed into a scanner.
Determination is a Choice
The following comes from Laura’s award-winning poster session for CAFA 98.She produced the written portion of her session using speech input (independently):
"My name is Laura Bridgeman. I am a 16-year-old student at Claremont School. I have been using voice recognition software for six months. It helps me read and write. I could not do that before because I have dyslexia. Having dyslexia means that when I see a word, it's just a word. It's just letters sitting on the paper. It's not saying anything to me. It has helped me read because it shows me the words.
In my life, I've been rewarded by sticking to things. In soccer I wouldn't be at the spot I’m in now if I didn't have determination. It is the same with my reading and writing. If I didn't have determination I still would not know how to read. It is rewarding because you get to do something on your own.
Determination is a choice. Some people choose to have and some don't. In the end, it is going to affect what type of person you are."
Name: Charlotte Tangley
Disability: Carpal tunnel syndrome
Computer: Pentium II Desktop 266Mhz 128 MB RAM
Dictating Environment: Cubicle in small office
Software: Naturally Speaking Preferred v3, Office 97
Roadblocks: Vocal nodules, high background noise in work environment
What made the difference: Vocal Hygiene and software friendlier to speech differences
Two years ago Charlotte was being counseled to start another career. Carpal Tunnel syndrome had made using a pencil, pen, or keyboard too painful. This pain put the technical writing she once did easily out of her reach. The promise of continuous speech recognition software was nothing less than a miracle.
Charlotte met five factors for successful use of continuous speech products:
2) intact speech and language system;
3) good working knowledge of the Windows95 environment and associated applications;
4) good initial training and support; and
5) a desire to write.
Initially she achieved good results with Naturally Speaking Preferred Version 2 (65 wpm and 96 percent accuracy). After a week she telephoned and told me that her accuracy was slowly getting worse. I noticed that her voice was huskier and asked if she had a cold or sore throat. She indicated that she was feeling well. I told her to go and see her doctor about her voice. I would see her next week.
As I suspected Charlotte had vocal nodules. During a second visit to Charlotte’s office, I used a sound level meter to determine the background sound level around her workstation. The average readings were 70 to 75 decibels (normal conversational speech is around 60 dB). She indicated that she was very conscientious in using her speech recognition software often going for an hour and 30 minutes before taking a cigarette break and drinking coffee at her desk. Charlotte needed to change her dictating environment and practices or she would be swapping one disability, carpal tunnel syndrome, for another, vocal nodules and perhaps loss of voice.
We found a quiet corner of the office where sound level averaged 50-55 dB. We instituted a vocal hygiene program that limited continuous use to 15 min with a five-minute break, water was substituted for coffee and smoking discouraged. Lower recognition levels served as a barometer of when she had to take vocal rest. With these practices in place and an upgrade to Naturally Speaking 3.0 Charlotte’s accuracy levels have returned to the high 90’s and her vocal nodules have subsided.
Just as Ray, Laura, and Charlotte are successful writers using voice input; other users with different disabilities can benefit from speech input. It’s not just an increased ability to write that can occur. Therapeutic advantages can be seen in users’ diction, intelligibility, breath control, stamina, and self-concept. The key to success lies in achieving an understanding of both the users' needs and an awareness of how the speech recognition application works. Modified training and commitment to learning can yield positive results in most situations.
As speech technologies move into the next century, they can change our classrooms, our workplaces, and the society in which we live. A person’s participation need no longer be limited by an inability to use and produce written language. New technologies will make the acquisition of literacy and the knowledge that goes with it possible for many with learning disabilities. The development of user-independent speech recognition programs will give more people with multiple disabilities access to the economy.
Doug Bowes is a special needs technology consultant for Special Education Technology - British Columbia (SET-BC), and can be reached at firstname.lastname@example.org or 2260 Victor Street, Victoria, BC, V8R4C5.