Imagine being able to turn up in any city anywhere in the world and speak freely with the locals.
It was fiction as Star Trek’s Universal Translator, in the hands of Lieutenant Uhura, chief communications officer on the USS Enterprise who kept Captain Kirk in touch with both Starfleet Command and the universe at large.
It’s now nearly a reality and may soon be in the hands of U.S. boots on the ground.
Raytheon’s BBN TransTalk is a two-way translation technology that can in real time translate explicit communication in languages like Arabic and Farsi into English.
The software runs on Android smartphones as well as Windows XP, meaning it supports most off the shelf laptops, tablets and desktops. A USB handset component combines microphone, speaker, and controls. And it clearly has a wide range of applications, including communication between military officers, force-on-force training and intelligence gathering.
In addition to being a useful tool in evidence gathering and interrogations, it could also be useful in tactical situations such as for patrols, gate guards, cordon and search missions, traffic control points, or even understanding or providing information over loudspeakers.
'If a human is able to easily read it, then [TransTalk] can do 60 to 70 percent accuracy.'
- Prem Natarajan, head of speech and language processing at Raytheon BBN Technologies
TransTalk also has tremendous potential for medical, refugee and aid assistance.
The technology was specifically designed to meet the requirements of the U.S. military and government, however, so you won’t find this excellent app in the Google Market. Raytheon will either send the software or pre-load it onto a smartphone. (Interest commercial customers can get in touch with the company for a price quote.)
In the field, soldiers, airmen, seamen and other fighters regularly come across foreign languages: road signs, graffiti, fliers, photographs, manuals, propaganda and more.
But a shortage of translators and the sheer volume of written foreign material means key information may be missed or analyzed too late to be of use.
Enter MADCAT. Multilingual Automatic Document Classification Analysis and Translation is a five-year technology evaluation program sponsored by DARPA to create technology that automatically converts foreign language text images into English.
BBN Technologies approached the challenge for Arabic by taking an image of the foreign language material with the handheld and processing it by a laptop on the spot.
Accuracy and speed depend on the amount of writing contained in the single image and whether it is handwritten or typed; printed matter can be read in seconds while hand writing takes a few minutes to decode. With more time and computer power, accuracy can also be improved.
The opposition can’t be expected to have A-plus penmanship, so legibility is an important factor. “
"If a human is able to easily read it then it can do 60 to 70 percent accuracy,” explained program director Prem Natarajan, head of speech and language processing at Raytheon BBN Technologies.
The problem isn’t easy: Beyond dialects and quality of writing samples, the wide range of writing implements is a factor, he said. Ballpoint pen or thick marker, it makes a difference.
In its fourth year of the five year DARPA contract, Raytheon has already made a French to English version commercially available.
The MADCAT program is just the beginning: BOLT represents the next step to putting truly universal translators in the hands of U.S. troops.
And BBN Technologies is already six months into cracking DARPA’s next challenge, the Broad Operational Language Translation (BOLT) program. SRI International was also awarded a $7.1 million contract for Phase I of the program on Monday, May 14.
BOLT focuses on understanding the implied meaning of words at a more advanced level. A successful technology would be able to ask and understand a response to open ended questions, such as “Do you have any concerns?”
In order to do that, the machine will need to be able to evaluate its own performance and recognize when it does not understand the meaning behind the words and ask the speaker questions for clarification.
Research will also develop understanding of new forms of social media such as Twitter that use language in a way difficult for a machine to understand.
The goal is a technology that will allow anyone to show up anywhere in the world and have a meaningful conversation with the locals. This once may have seemed far-fetched. Researchers get closer every day.
Ballet dancer turned defense specialist Allison Barrie has travelled around the world covering the military, terrorism, weapons advancements and life on the front line. You can reach her at firstname.lastname@example.org or follow her on Twitter @Allison_Barrie