Speech "re-cognition"? Investigating speech as a cognition-augmenting modality

We report and compare two projects investigating speech interfaces and augmented cognition. We performed an exploratory study to examine the effects of speech-enabled input on an imagery analysis and annotation task. We hypothesized that speech recognition could be a cognition-enabling technology by reducing the mental load of instrument manipulation and freeing up resources. Results indicate that people did identify images more efficiently and could potentially annotate images faster with speech. However, people did not annotate better with speech (precision was lower, and recall was significantly lower). We also developed a personal digital assistant capable of controlling a search-and-rescue robot. The PDA displayed a map and video feed of the robot´s camera and permitted both stylus and speech input. We note differences between speech input in the robot control task and in the imagery annotation experiment. We conclude that speech may be helpful if the cognitive cost of speech recognition delays and errors is outweighed by the cognitive benefit of reducing instrument manipulation.