Cutting-edge voice, speech and translation technology from IBM is getting a critical test from U.S. military personnel who are using the system to speak with Iraqi locals in their own language.
The Multilingual Automatic Speech-to-Speech Translator, or “MASTOR,” is activated by speaking and translates both from English to Arabic and from Arabic to English, sounding out its translation instantly in audible speech.
The U.S. Joint Forces Command (USJFCOM) is addressing a shortage of military linguists and putting the technology from IBM Research to work in the field. The software is compatible with PCs, laptops, tablet computers and personal digital assistants.
The U.S. will deploy the speech-to-speech translation system in Iraq on 35 ruggedized laptops for the Army Medical Department, Special Operations Command, the Marines and other Department of Defense units.
“Our goal is to enable units operating in areas where human interpreters are scarce to communicate effectively with speakers of different languages in real-world tactical situations,” said U.S. Joint Forces Capabilities Division Branch Chief Wayne Richards.
Executives behind the new MASTOR technology compared it to other translation solutions that rely on preprogrammed, fixed phrases, indicating the IBM software “offers the ability to have a free-form conversation without having to memorize any predetermined phrases.”
MASTOR is designed to take spoken input from its users and convey the meaning of what is said in translated language, which initially includes Iraqi Arabic, Modern Standard Arabic and Mandarin Chinese.
“The military’s use of the MASTOR system is a very exciting example of that capability — one where we see the potential to improve the safety of U.S. Service personnel and save lives,” said IBM U.S. Federal Managing Director Anne Altman.
The technology represents a significant leap in translation and communication systems, according to Basex Chief Analyst and CEO Jonathan Spira, who has yet to view a demonstration of the project.
“MASTOR promises free-form conversation, which is how people actually speak,” Spira told TechNewsWorld. “This is a huge advance in technology compared to systems which only can handle predefined phrases.”
MASTOR, like human translators, overlooks minor errors, he noted.
Bridging the Gap
Rather than relying on a specific device, MASTOR was designed and optimized for multiple platforms and runs on Windows XP, Windows CE, Linux and Embedded Linux, IBM Research Manager for Speech Recognition and Understanding Yuqing Gao told TechNewsWorld.
The technology’s potential is broad. It can be incorporated in a variety of different software applications and adapted for other languages, which IBM is already planning to do, she said.
For U.S. military personnel in Iraq, MASTOR has been tuned not only for medical, infrastructure and other communications, but also to avoid conflict, according to Gao.
“A lot of conflict is caused by language barriers,” she said. “We’re trying to overcome that.”