Google Translate has become better because of neural networks
Application of neural networks has allowed Google to improve the quality of translations by Google Translate. Now the app works more accurate and better conveys the meaning of the text even when using the offline version. It is reported by Cult of Mac.
For many years the possibility of neuro-machine translation helped the online version of Google Translate is becoming better. In June 2018, the functions of artificial intelligence was added in the offline application that can run on iOS and Android offline.
The more often the user accesses the service, the more natural it becomes and the better the translation. Since the application does not need network connection, it can be used even in travel. The user does not need to seek the Internet to ask for directions or understand what is written.
“Neural system translates sentences as a whole, not in parts. It uses a broader context, to help identify more relevant translation, which is then reconstructed to sound as properly speaking a real person. This makes the translated paragraphs and the article more clear and easy to read”, — the statement says Google.
Google Translate is not the only service that use achieve artificial intelligence. In the beginning of 2018 Microsoft Translator also got a neural network that can operate in standalone mode.
To work with an improved version of Google Translate will need to install vocabulary packs for each of the used languages. Dictionary sizes range from 35 to 45 MB, which is much less than solutions for traditional machine translation.
Updated version for iOS will appear in the App Store soon. The Android app can be downloaded from Google Play.
Follow the news in our Telegram-channel, as well as in the app on iOS MacDigger.