4. What exactly constitutes a voice recognition system?

A voice recognition system is made up of a computer with system software, voice recognition software, a microphone, and usually a sound card. To use voice recognition to word process, a word processing program is also needed. Each software program has different hardware requirements, but generally speaking a more powerful computer is needed--typically with a Pentium or a very fast 486-based CPU and at least 16 MB of RAM.

In general, the voice recognition software itself is built on three parts: a large electronic dictionary (e.g., a 150,000 word dictionary from some publisher such as Merriam-Webster), a smaller active dictionary that reflects the user's own usage, and a voice model.