Tag: Neural Machine Translation

Google has made a big change behind the scenes with Google Translate that is going to dramatically increase the accuracy of translations within the app.  In September, Google announced they had developed their Neural Machine Translation and discussed it in a bit more detail at the October event.  Now they have begun rolling it out to Translate with support 8 of the 130 languages that can be translated with the app and service.  This is a first step but the implications are huge simply because of now Neural Machine Translation works.

Today, Google Translate chops up whatever you are trying to translate and converts that into the foreign language.  It is good – really good – but it could be better as it doesn’t necessarily know local idioms and other speech traits.  Neural Machine Translation essentially looks at the entire sentence, looking at context and then coming up with a translation that more closely matches human speech instead of a machine.  It means that when you use the app it will give you far more natural results.  Equally as important, the app will learn as it goes to provide more accurate and better translations as time goes on and more people use it.