As widely reported yesterday, Facebook has acquired Mobile Technologies (MT) and its translation app “Jibbigo.” There are a wide range of potential applications of the technology for a global site and platform such as Facebook. They extend to translating real-time communications in Messenger, translating News Feed and even ads.
Facebook’s long-term vision for MT probably goes well beyond these relatively straightforward translation scenarios. In a sense this is another “Facebook is a mobile company” play. But it’s far from mobile only. The technology can be used in PC and “offline” contexts as well.
Facebook likely bought MT at least as much for the people behind it as for the tech. However there is real technology here. It’s not an “acqu-hire.”
Jibbigo released the following statement on its website yesterday:
Mobile Technologies was founded in 2001 with the goal of breaking through language barriers to open up communication between the people of the world. With 6,000 languages in the world, the language divide presents the greatest human communication challenge. Mobile Technologies has developed a number of cross-lingual communication tools: Jibbigo, the world’s first speech-to-speech translator on a phone that runs online and even off-line, independent from the Internet. Travelers around the world use Jibbigo to communicate in foreign countries and health-care workers overcome the language challenge in humanitarian missions. Mobile Technologies also developed and deployed the first automatic, simultaneous interpretation service for lectures and deployed it in educational settings, so that ideas can transcend nations and cultures.
MT was described in some articles as a “speech recognition” company. However my understanding is that there is no “speech recognition” in the sense we usually mean that term. MT and Jibbigo chiefly rely on Nuance speech recognition capabilities. The technology that Facebook is buying is machine language translation.
The noteworthy dimension of Jibbigo and MT, as the statement above indicates, is that there’s no server or internet connection required for the app and the technology to work, although MT does offer server-based translation as well. In a demonstration of the latter, in June of last year the company instituted a lecture translation program in conjunction with Karlsruhe Institute of Technology in Germany in which German language lectures were translated in real time into English (subtitles on iPads and video screens).
Prior to the acquisition MT was seeking to expand those capabilities more broadly into university scenarios and business contexts (webinars, conferences, etc.).
MT is one of the stronger independent providers of machine translation (assume Facebook did its due diligence). However there aren’t many out there.
Machine language translation is at once amazing and not all that good currently, especially compared with human translators. Believe it or not, the field is dominated by Google and Microsoft; most mobile translation apps rely on one of the two. Now Facebook could further advance the technology in interesting ways and even make it more broadly available to developers and third parties.