Comparison of Speech with Keyboard and Mouse as the Text Entry Method

Abstract

Federal Aviation Administration (FAA) researchers ran an experiment that compared an automatic speech recognition system with the keyboard and mouse as text input methods. In the speech condition, the participants read the document in its entirety, and the system converted speech into text. They corrected errors using the keyboard and mouse. In the typing condition, the participants corrected errors as they typed. The experimental results showed they spent significantly less time reading than typing to enter the text. When we factored in correction time for both conditions, the results showed that participants took significantly more time in the speech than in the typing condition. Whether they were fast or slow typists, all of the participants preferred typing to speech and performed better in the typing condition than in the speech condition. Optimization of the system and more training may improve the performance of the speech recognition system.