menu

News

Can artificial intelligence replace a translator’s brain?

posted in Uncategorized by

The ties between translation and technology are a matter of great interest for STUDIOTRE.
Technology plays a key part in the work of translators. CAT Tools and Translation Memories (TMs) are tools of the trade that we use on a daily basis. Meanwhile, machine translation plays no part in our everyday working lives but it is an area that has seen a lot of experimentation in recent years.

Last year, we wrote on our Facebook page about Google Pixel Buds: Bluetooth earbuds that understand what you’re saying and translate it into one of a selection of 40 languages in just a few seconds. A few weeks ago, Google was at it again. This time, it presented a new machine translation system called Translatotron.

What’s so innovative about Translatotron?

There are normally three stages in the machine translation process:
1) Voice recognition technology is used to transcribe the spoken words from the source language as a written text.
2) The written text is translated into a written text in the target language.
3) The written text is summarized vocally.

In contrast, Translatotron is capable of going straight from the original spoken words to the translated spoken words in one step (skipping the written texts) while also producing the same voice and tone as the “real” speaker, thus doing away with the sound of the machine translation system. In jargon, this is known as speech-to-speech translation.

So will artificial intelligence ever be able to replace the brain of a translator or interpreter?
These innovations can unquestionably help to break down language barriers between people in certain circumstances, but when localization and transcreation are required in the translation process, human input is essential.

At present, machine translation systems are unable to distinguish between the different styles and purposes of texts. Furthermore, they can easily be tripped up by idioms, the rhythm of sentences, metaphors and all references to specific languages, cultures and outlooks on life.
For example, Google Translate incorrectly identifies Sicilian as Corsican, while Bing Translator gets Lombard mixed up with Catalan.

Have you tried out the innovative new systems in the world of machine translation?
Get in touch with STUDIOTRE to let us know how you got on!

03 Jul, 19

 

 

See also