Published in News

Researchers develop software to extract meaning from speech

by on26 January 2010


Applying pragmatics to man-machine interaction

A group of forward-thinking researchers at American aerospace and defense contractor Lockheed Martin are currently exploring software methods that can extract meaning and context from a string of natural human conversation. In other words, the linguistic subfield of pragmatics is now being applied to computation and might just lead to the emergence of an era where fictional Star Trek man-machine conversations might surface into reality. But as the story usually goes, these technologies are always kept in the confines of government hands until deemed appropriate for average consumer deployment.

Nevertheless, the software technology in question is known as SPLICE, a rather intriguing abbreviation for Spoken Language Interaction for Computing Environments. In perspective, the significance of the development is largely attributed to its ability of interacting on a much deeper cognitive level with humans through natural speech. According to Kenny Sharma, an engineer at Lockheed Martin Corp.’s Advanced Technology Laboratories, the software can identify a speaker's intent and draw logical conclusions based on the information available in context.

The software is extremely flexible and can be applied to a broad range of industries. However, one of the prominent uses for SPLICE technology is in battlefield medicine. The idea is that the development of a medical voice documentation system for trauma patients during wartime would significantly reduce ambiguities in patient information as details about trauma cases get passed along the chain of medical intervention in war zones. Rather than spending precious time-critical minutes gathering data about fatally injured patients, the system audio records medical personnel as they speak during treatment.

“If we can passively capture that information, it’s really valuable stuff,” says Sharma, a member of the Lockheed Martin team developing the prototype.

This is just one of the various practical uses for the developing software technology that Lockheed Martin researchers have in mind. In the consumer space, the possibilities are nearly endless. Google’s voice recognition technologies have taken the mobile market by force with the inclusion of voice-powered search applications for iPhone and Android OS devices, not to mention its audio-indexing for political documentation on YouTube. Combine the power of location-aware GPS technology with voice recognition and some context-aware SPLICE and it might someday be possible for your mobile device to identify some of your rudimentary intentions.

Last modified on 26 January 2010
Rate this item
(0 votes)