Last November 2016, Google introduced Neural Machine Translation (NMT) technology. It is an artificial intelligence-based interpreter technology to deliver more accurate, natural, and easy-to-understand translations. The good news, starting this week, the technology is available in Google Translate application for Bahasa Indonesia.
Improving the accuracy of the translation can not be separated from the studies that have been done. Based on research by Google Translate team in Indonesia, there are at least 5 things that become preference from Indonesian users when using an online translator. That is easy to use, fast translation process, can work between devices, has audio playback feature to know how to pronunciation, and no less important, grammar. This NMT technology will see the whole sentence (no longer words per word), using the broader context to find the most relevant translation results, tailored to the language of the conversation with the right grammar.
Intelligently, this NMT technology will see the whole sentence, no longer words per word as before which is confusing when read. The technology also uses a broader context to find the most relevant translation results, as well as tailored to the correct language of conversation and grammar.
Macduff Hughes, Engineering Director, Google Translate, says Indonesia is one of the most active Google Translate users in the world. “So we are excited to be launching this technology for Indonesian Neural Machine Translation will help deliver better and more natural translation results,” Hughes said.
Currently, NMT technology in Google Translate is available for English to Indonesian translation, and vice versa. Outside Bahasa Indonesia, Neural Machine Translation is also applied to 40 other languages (from and to English), such as Korea, Japan, China, France, Germany, Spain, and Russia.
This update can already be used in Google Translate, both through Android and iOS apps, desktop (translate.google.com, and automatic translator feature in Google Chrome. Good job, Google!