In Douglas Adams’s humorous sci-fi novel series Hitchhiker’s Guide to the Galaxy a special kind of fish is mentioned — the Babel Fish. When inserted into the ear, it translates any spoken language into whichever language the listener understands. It is a very nifty device, and now Google seeks to create something similar.
According to Times Online, Google is developing a speech-to-speech automated translator for Android phones. It’s essentially a combination of two of Google’s existing technologies; its online universal translator service, Google Translate, and its voice recognition system.
Google plans to make its Babel Fish a lot like a human translator; the software would analyze chunks of speech, and translate them in their entirety rather than translating word for word. Franz Och, Google’s head of translation services, claims the technology could go live in a couple of years. “Clearly, for it to work smoothly, you need a combination of high-accuracy machine translation and high-accuracy voice recognition, and that’s what we’re working on. If you look at the progress in machine translation and corresponding advances in voice recognition, there has been huge progress recently,” he says.
Anyone who’s used Google Translate knows that translations aren’t (and probably never will be) perfect, but they’re very helpful when you can’t understand a word of some foreign language. However, Google’s voice recognition also has issues of its own, and I fear that these two combined would produce a very high amount of errors. The Times also mentions the issue of different accents, a problem that Google plans to solve by making the software gradually learn the speaking habits of the phone’s owner.
Despite the big issues Google has to overcome to make this technology useful, if anyone can pull it off, Google can. The idea of being able to call someone who doesn’t speak your language, and have the conversation translated almost instantly, well, that’s one of those inventions that made Google the company it is today.
Tags: Google, machine translation, translation
