"The sounds entering the microphone pointed away from the mouth are then subtracted by the software," Black said. "This method works surprisingly well to filter out vehicle noise or to eliminate the babble in a crowd of people."

Black said he agrees with M*Modal's approach of encouraging editing of the machine-generated text by a skilled transcriptionist and preserving the text and audio recording in cases when that is impossible.

The approach taken by Think-A-Move in its TATRC-sponsored project was to deploy a headset that is able to capture sound in the ear canal. "When a person speaks, sound travels back into the ear canal," Brown said. "The sound in the ear canal includes significantly less outside noise and the headset has an earbud microphone that provides a high level of recognition accuracy."

Software that Think-A-Move developed for this project uses algorithms to filter out ambient noise from human speech. Think-A-Move's earlier research included using speech commands to control radio communications in aircraft and in tactical military ground vehicles.

The Starix/Think-A-Move project is proceeding apace, according to Battaglia, with a series of operational demonstrations to be concluded later this year. At that point it will be up to one of the Army's program offices to pick up the project and develop it into a product that could be use on the battlefield.

Battaglia's view squares with that of Geesey, who estimated that a prototype system could be ready later this year but that it would take two more years – probably longer – to deploy a voice recognition system to combat medics.

Further evaluations of the developing technologies are slated for fiscal year 2012, according to Gilbert, with one major hitch: Funding for these projects has yet to be approved.