Input devices: a usage-driven approach


Here are some information about how I put my work environment together. For a picture of my workspace, see User interface summary.

Foot pedals

It is difficult, if not impossible, to buy appropriate foot keyboards. Foot switches that just float around on the floor are a bad idea. Your feet won't be able to find them—and using them frequently may induce cramps in calf muscles. Foot switches should be mounted on a foot rest. A Y-key dual keyboard adaptor (from allows two keyboards to be used together. Also, I believe that USB keyboards can be used in parallel. In any case, individual foot switches can connected to the soldering points under each key on a second keyboard. It's easiest to work with a keyboard equipped with mechanical switches, since capacitive switches may require the addition of a capacitor for each key. The construction of a keyboard like the one shown probably can be done in a couple of days by somebody with electronic and woodworking skills.

Buttonless mice

Under Windows 2000 and XP, it is easy to make several mice work together. Personally, my favorite "ergonomic" mouse is a 10 dollar device from Sakar International (Model #77452, Iconcepts). It is small so that the hand does not have to struggle mounting it, and it is not too rounded. I have disabled the mouse buttons by sliding two thin slices of rubber in the gap between the button and the case. A better alternative is to reshape an optical mouse, by peeling off the upper casing; a better surface for grasping the mouse can then be built out of epoxy or modeling clay. Electrically, an ordinary foot switch or an extra key mounted on the keyboard over a micro switch may be used to short-circuit the two contact points of the micro switch serving a button inside the mouse. Thus, with relatively modest soldering skills, one can connect a mouse button switch via a second mouse (that is used solely for this purpose). Under Windows 2000 and XP, it is possible to press the button of one mouse, while moving the other mouse, to achieve the effect of dragging. This is not possible under some other operating systems, including Windows 98 and Me.

There might be easier ways to obtain a buttonless mouse—I have not investigated alternatives.

Wireless microphone

I like the design of the wired Emkay D-loop microphone. It is very comfortable to wear, and the microphone stays in place. Unfortunately, I have not been too impressed with the wireless consumer solutions that I've tested. Instead, I use a Shure LX wireless system. The microphones that Shure provides come with elastic bands that wrap around your neck. This a great solution for aerobics instructors—but an unworkable solution for speech recognition: you can't move your head without the microphone moving in front of your mouth. Instead, I have connected an Emkay D-loop microphone to the mini-XLR connector. The details are: connect shield to pin 1 and short-circuit pin 3 and pin 2 (which carries 5 V) and connect the hot wire (red) or pair of hot wires (red and white) to that short (inspiration for this operation is the diagram found at To accomplish this wiring, I had to open the female mini-XLR connector. To my knowledge, there is no mini-XLR to mini phone plug connector on the market.


I used to use an IBM keyboard without numeric keypad to gain space. This excellent keyboard has mechanical switches and widely spaced Shft, Ctrl and Alt keys (IBM part no 1397681). It is not made anymore. My current keyboard is different, but since finally I am able to largely not use it, its design is not important.


In my experience, Pentium III processors under 1 GHz are inadequate for speech recognition based on the NaturallySpeaking dictation system, now produced by Scansoft. It seems like some of some kind of “thrashing” (a technical term of computer architecture) frequently occurs when a response to even a short utterance takes several seconds (perhaps, the L2 cache of the CPU is too small.) The last six months I have been using a AMD-based processor (Athlon 1.4 GHz) with extremely nice results: response time for short utterances is very quick and the system altogether has been very stable, even when slightly over-clocked (something that is unnecessary). I have no experience using Pentium IV systems.