Following the September 11, 2001 attacks in the United States, the U.S. and British forces have resided in Afghanistan. During the past decade the military relied on human translators for communicating with non-English speakers, but due to the dangerous nature of the job and the lack of truly skilled translators, the pursuit of alternatives became necessary.
To the request of the Defense Advanced Research Projects Agency (DARPA), a team of scientists from the National Institute of Standards and Technology (NIST) designed a translation system in use for the past four years. The system, dubbed TRANSTAC (spoken language communication and TRANSlation system for TACtical use) aids translating foreign languages to English and vice versa. Unlike previous devices – that used microphones and portable computers – this one is more compact: Tt is based on smartphone technology, making it as small as a bulky cellphone.
The development process involved evaluating three two-way, real-time, voice-translation devices designed to improve communications between the U.S. military and non-English speakers in foreign countries. Currently, the focus is on Pashto, a native Afghani tongue, but NIST has also assessed machine translation systems for Dari – also spoken in Afghanistan – and Iraqi Arabic.
According to the project’s manager, Craig Schlenoff, the system works as follows: An English speaker talks into the phone. Automatic speech recognition distinguishes what is said and generates a text file that software translates to the target language. Text-to-speech technology converts the resulting text file into an oral response in the foreign language. The same process is used for the foreign language speaker.
In order to determine critical communication interactions to simulate and evaluate in tests, the team of NIST researchers held focus groups with U.S. military personnel. Then, they devised 25 scenarios for evaluating the performance of translation devices, including vehicle checkpoints, communication of key information (regarding utilities such as electricity and water), and medical assessments.
The goal was to simulate these situations as realistically as possible; therefore, the U.S. Marines and native Kandahari-dialect Pashto speakers acted out the scenarios without a script. Each scenario was performed using the three translation systems. For each test, on-site judges observed the scenarios, and the participating Marines and Pashto speakers were surveyed about the ease of interaction with the systems. Later, a separate panel of judges fluent in English and Pashto viewed videos of the exercise (see this YouTube video for demonstration).
The videos were reviewed, and each one of the three systems was scored in terms of how accurately concepts were communicated in both languages. “We are writing a detailed assessment of the evaluation for DARPA so they can make an informed decision to determine where to direct funds and efforts in the TRANSTAC project,” says Schlenoff. Hopefully, the finalized system will support multiple languages.
TFOT has also covered DARPA’s new high resolution camera, which has 2.3-gigapixel quality, and DARPA’s Surveillance Blimp, designed to fly higher and longer than any existing intelligence, surveillance, and reconnaissance aircraft. Another related TFOT story is that of Fuji Xerox’s Copy-and-Translate Machine, capable of translating scanned Japanese text documents into Chinese, Korean, and English.
For more information about NIST’s translation system, see the formal press release.