Google, Skype Race To Tear Down Language Barriers

It feels like an attempt to reverse the effects or even create a new Tower of Babel, but this time no actual towers are involved (unless you count cell towers). The bridging of languages is, instead, being done over the Internet, in real time, and using your voice. And at the forefront of these advancements in technology are Google and Skype (and by proxy, Microsoft) who are now starting to compete not for the best voice chat service, but for which one can translate a spoken word better.

Before the year ended, Skype announced its new Translator preview that translated languages as they happened, like in a conversation between two people of different tongues. At the moment, only English and Spanish are supported and the feature is limited to a few testers. But it's not hard to imagine this feature catching on quickly, which will have Skype scrambling to support more languages faster.

Google, however, won't be outdone. Now there is talk that the search giant has an update to its translation services that rolls in that feature as well. Google Translate is already quite talented, supporting over 90 languages. It can also already identify some spoken words in another language, but the new real-time identification and translation is definitely a step up from the status quo.

It may all seem wonderful, magical, and definitely helpful, but it's not without cost, at least in terms of potential costs to privacy. These translation services aren't simply fed translations word for word and then left on their own. The early days of services like Babelfish are testament to how that can end up really badly. The complexity of human languages require these systems to continually learn, and even for that kind of machine learning, you will need data, lots of data and constant data. That is why these services usually have some disclaimer or user that asks if you'd like your words to be used to help improve the system. Of course, that also means that they would store your conversations, which can give rise to privacy concerns.

Naturally, these companies insist that user data is anonymized and the users are protected. Skype says that conversations are broken up into files and lose their connection to a particular sender or speaker that even the NSA won't be able to make sense of it. Google even takes obfuscation one step further. Because of the possibility of voice being used for biometric-based security like passwords, the voice used for the translation is different from the user's own voice and gender, which, of course, sometimes make for pretty amusing situations.

SOURCE: NY Times