A merger of two companies on Tuesday, Dial Directions in San Francisco and Sakhr Software Co. in Vienna, Va., brings together technologies that can turn an iPhone, BlackBerry or Windows Mobile device into a voice and text translator.
Dial Directions has developed voice activation and recognition software that is used to provide driving directions via cell phones. Sakhr makes Arabic speech and language technology for U.S. government agencies, among other customers. The companies have been working on a joint project before the merger, but are now discussing how it works.
The new application is a mobile voice and text translation application that enables a user to speak English or Arabic into a phone. That speech is then translated into speech and text via a cloud-based service that comprised the application. It's not a real time universal translator, but it's very close.
"You can say anything you want and it will translate it very, very well," said Dial Directions CEO Adeeb Shanaa, who is now chief executive of the combined companies. "I would put it close to very close to actually having a perfectly fluent translator with you in your pocket."
Voice training by the user isn't needed, but the voice recognition will improve after a few times of use, said Shanaa.
Dan Miller, an analyst at Opus Research, which studies mobile Web and speech processing, says the merged technology's capability is "something of leap."
While he hasn't tried this specific device, Miller said translation capability, as general rule, is still short of perfect. If able to achieve between 70 and 80 percent accuracy transcribing a single language into to text, "you are doing very well."
The U.S. government is seeking 95 per cent accuracy in the translation of Arabic and Chinese speech and text under a research program it calls Global Autonomous Language Exploitation (GALE), which is being run by the U.S. Defense Advanced Research Projects Agency (DARPA). Sakhr is involved in the GALE project.
All the Dial Directions employees have joined Sakhr, which now has a combined workforce of 200. Shanaa said the basic technology can be applied to other languages.
