Google Translate has changed the work atmosphere of most of the translation service provders in the last one decade but what Google, for that matter, others too have failed was to provide a comprehensive sentence-wise and intelligent conversion of phrases and local sayings into equally sounding sentences in other languages.
Last year Google introduced its neural machine translation to improve its Google Translate tool based on statistical models to translate text. Unlike the traditional statistical machine translation, the new machine translation translates complete sentences at a time, instead of pieces of a sentence. It is like "more human sounding", said Google in a blog post.
Apart from English, Google has taken up the new translation to the doorsteps of those speaking or using nine Indian languages, including Hindi, Bengali, Marathi, Gujarati, Punjabi, Tamil, Telugu, Malayalam and Kannada. Currently, 29 languages are enabled with neural machine translations in Chrome browser and Google promises to add more in the future.
Explaining the sentence-to-sentence translation, Google in its blog said the new software first rearranges it and then adjusts it to sound more like a human, and then offers a solution the way it had learnt over a period of time to offer translations that are easier to read.
Google’s neural machine translation system was developed by using a human side-by-side evaluation on a set of simple sentences and reducing errors by about 60%, compared to the phrase-based translation system. Google has also made the Rajpal & Sons Hindi dictionary available online.