Themes | Papers | Clairgrove | My workspace | CV

My workspace: speech recognition and foot keyboard

My computer workspace is an unusual one. It is designed with the goal of virtually eliminating the usual physical strain on fingers, hands, and arms. Therefore, the use of these body parts must be drastically reduced—something that no amount of tinkering with keyboard or mouse design can achieve. The other goal is that the reassigned workspace be more efficient the conventional one; otherwise, there is no incentive to start working in different ways. These goals are usually regarded as irreconcilable. Attempting to solve the problem, I have been motivated by a typing injury (for that story see CTD-RSI).

After ten years of trial and error, I believe that I have found a promising approach. The solution is not commercially available. And, I don't know to which extent my ideas apply in general for the many people who are unable to use the keyboard and mouse in the intensive way necessitated by modern jobs.

My low-impact workspace concept integrates three user interface ideas:

  • replace the keyboard with speech recognition for dictation and editing;
  • defer repetitive work to foot pedals; and
  • replace the mouse with a buttonless pointing device that is pushed around on a large surface.

The physical expression of this concept can be seen in this picture:

Photo of workspace, showing rest area for hands,    buttonless mouse, and foot pedals.
Click here for larger picture.

The video demos demonstrate how composing a letter is a breeze with speech recognition and how repetitive tasks are accomplished easily by foot work.

  • The video ShortTalk for composing a business letter (streaming AVI format) [mpeg-1 version (12Mb)] shows how speech recognition serves two roles efficiently: that of automated transcription into a text editor and that of editing the resulting text through the use of ShortTalk. The editing commands of ShortTalk are those words that sound unusual: they must be different from English, since that is the only way to fluently mix dictation and commands (otherwise, how would the machine know what is what?). ShortTalk is several times more efficient than the so-called natural language commands of commercial dictation systems.
  • The video foot keyboard (WMW, streaming Windows Media) [mpeg-1 version (28Mb)] shows how all the keys that are commonly used for repetitive tasks can be activated by feet through a foot rest surrounded by switches. The foot keyboard effectively complements speech recognition, since it relieves the voice of tiring work for which it is not suited. For more information, see Input devices: a usage-driven approach.

With the techniques illustrated in the videos, it is possible to work while leaning backwards in a relaxed manner, since hand work and finger work has been reduced to just occasionally pushing the mouse around. Therefore, the workplace puts less stress on neck and back as well.