Experts translate voice tech needs

The universal translator made popular in Star Trek movies is still years, if not decades away, but speech and voice recognition technologies can — and are — helping the U.S. armed services achieve their missions.

However, government users must better define their requirements to get applicable tools into warfighters' hands, according to government and industry experts.

"Our responsibility [in the armed services] is to better explain what we need to do," said Ashley Johnson, science adviser for the Marine Corps' Marine Forces Pacific. "It's the art of the possible with requirements. For high-tech analysis and intelligence stuff at the three-letter agencies, there are good tools, but moving that capability to the front line targeted to mission areas, we're just starting to scratch the surface."

Speaking July 30 at a speech and voice technology conference in Washington, D.C., Johnson said getting quality technologies, including translation tools, into warfighters' hands requires a "management of expectations" from Defense Department users and commercial technology providers.

He added that pocket "phrase-a-lators," which can translate basic phrases in foreign languages, are being used in Afghanistan and elsewhere in the world with some success.

Judith Markowitz, president of J. Markowitz, Consultants, said the speech and voice technology market is expected to be worth about $6 billion by 2006, with government and military customers accounting for 15 percent to 20 percent of that spending.

The three main types of technologies are:

* Voice biometrics, which includes speaker authentication.

* Speech and voice generation, which includes text-to-speech synthesis.

* Speech and language recognition, which includes audio mining.

But if a vendor suggests a device is "all things to all people," military users will quickly be disappointed when it doesn't meet expectations, and then they will "throw it in the back of the truck," Johnson said.

"You have got to target a device towards an environment," Johnson said. "If a Marine understands what he's got and how to use it, even if it's only 70 or 80 percent, he can deal with that."

Jake Hodges, vice president of operations at Eumetria Inc., a consulting firm that has evaluated numerous speech and voice technologies for government agencies, agreed and said that determining operational requirements and how a device is going to be used are the key elements.

"First, you have to understand what a users' requirements are or they will take the systems out there and the portions that it applies to [will use it] and for the others that it doesn't work, it becomes a doorstop," Hodges said. "One piece of equipment can't satisfy everybody."

However, dependability is equally important as a small footprint, reliable power source and usability, Johnson said.

"It must also be user-friendly," he said. "If you need a Ph.D. to use it and you hand that to a 19-year-old Marine who is used to five or six sentences explaining how to use [equipment], that's not the right environment." But that tool could work for a senior intelligence analyst, he added. "You have to understand the target."


  • FCW Perspectives
    zero trust network

    Why zero trust is having a moment

    Improved technologies and growing threats have agencies actively pursuing dynamic and context-driven security.

  • Workforce
    online collaboration (elenabsl/

    Federal employee job satisfaction climbed during pandemic

    The survey documents the rapid change to teleworking postures in government under the COVID-19 pandemic.

Stay Connected