After 10 years since its launch, when it revolutionized the translation software market, this week the Google translator finally got a significant update.
Until now, Google Translate was only accurate with isolated words, and what we saw in sentence translations were confusing and inconsistent results. This was because the system always translated words individually and only then organized them into a meaningful form, which in practice didn't actually happen. But with the new update, this problem may have been resolved.
With the new update, Translator has gained neural networks, a feature that allows Google to decipher entire sentences at once. This makes the translation more fluent and more faithful to the original sentence, mainly because neural networks enable reading that takes the content's context into account. Translator reads sentences and presents the result based on the information contained in the entire paragraph, which was not the case in the previous version.
According to Google, this change in reading method makes results much more natural for native speakers of certain languages, and with it, translation errors should initially decrease by up to 60%. This number should reduce even further over time, as neural networks continuously improve based on user interactions.
The new Google Translate is now available on the company's official website and in apps for Android and iOS, but for now it only supports eight languages: Portuguese, English, French, German, Spanish, Chinese, Japanese, Korean, and Turkish. This represents only 35% of the system's 103 supported languages.
The new feature is expected to be released for all other languages soon.
