Speech Recognition Considerations
To consider speech recognition as assistive technology helps to use a process. Here, the variables of the SETT Framework (Zabala. 1995) are used to create a task-specific consideration process (see Socol Toolbelt Theory):
1. Consider the TASK: Are the writing demands significant? Expectations vary from region to region, but most standards currently DO require a lot of writing, even in the younger grades.
2. Consider the ENVIRONMENT: Would the use of speech recognition match the environment?
-
- Yes, when talking out loud would not be disruptive. If NOT, the assignment may need to be finished at home or in a quiet location at school (the back of the classroom actually works fine for some students).
- Yes, when environmental noise does not affect recognition accuracy. A noise-canceling microphone can filter out talking, but a lot of surrounding noise makes recognition worse.
- Yes, when talking out loud would not be disruptive. If NOT, the assignment may need to be finished at home or in a quiet location at school (the back of the classroom actually works fine for some students).
3. Consider the STUDENT'S SKILLS: Is there a gap between the student's current writing skills and curricular expectations? Can the student learn the skills needed to functionally use speech recognition?
Generally, speech recognition works well for students, who can express thoughts and ideas better verbally than in writing. It has a good chance of working if the student can learn or already has the following skills:
Generally, speech recognition works well for students, who can express thoughts and ideas better verbally than in writing. It has a good chance of working if the student can learn or already has the following skills:
- The consistency of enunciation, volume, & pitch
- Syntactical speech patterns (understanding of grammar)
- Ability to inhibit "uhms" and "ahhs"
- Ability to express ideas in verbal language
- Attention to auditory and visual details
- Ability to multitask
- Ability to edit
- Ability to problem-solve
- Ability to self-monitor
- Ability to tolerate frustration
- Motivation to use technology as an alternative writing method
Understanding the Speech Recognition Writing Process
This process requires the integration of many underlying skill sets:
- Language
- Generation of ideas
- Short-term memory
- Oral articulation
- Fine motor
- Decoding
- Editing & revising
- Computer operation
Writing with speech recognition software is different from handwriting or typing and requires the integration of a different set of skills.
Here is a task analysis of the speech recognition writing process.
1. THINK about your topic and what you want to write.
2. Mentally compose your sentence.
-
-
- You may need to talk it through (silently or out loud) to get it to sound right.
- Decide which punctuation is needed at the end and whether any punctuation is needed in the middle (e.g. commas. quotation marks).
- Hold the sentence (or part of it) in memory.
-
3. Turn on the microphone.
4. SAY your sentence with clear enunciation but in a natural manner (not too slow or too fast). Verbally add the correct punctuation at the end.
5. Turn off the microphone.
6. CHECK the recognition accuracy. Students with weak decoding skills may have difficulty with the editing process.
7. FIX the recognition errors manually or verbally.
8. Repeat the process for the next sentence.
