Machine Translation

Neural Machine 

Machine translation (MT) has developed a bad reputation until recently, because it was more distracting than helpful to use MT-assisted translation memory (TM). Neural Machine Translation (NMT) is a new technological jump and seems to transform our understanding with amazing results to a hope for new era. 


Using carefully controlled authoring, customized MT engines with the most up-to-date glossaries and memories, and human post-editing, MT is now reaching to a stage where it can produce TM quality output faster than TM itself for many types of texts. In general, MT costs will set in between TEP rate and proofreading while productivity get even better when volume is large enough.

Neural Machine translation is a future, as it is fed by rising amount of human-translated bilingual content. Since the first introduction of Trados in 1990s, translation memories have been with us. Today these translation memories represent the huge amount of input to Neural Machine Translation engines. 



Depending on the quality levels of the source, NMT may work well. Our dedicated human post-editors compare the source and target segments to establish accuracy, consistency and tone of voice. 


Extra-large content

NMT engines are able to learn from old TMs, which makes it possible to translate the large contents by running NMT engines and post-editing by human post-editors. 


Project management

Project management, terminology management and language quality assessment must be carefully taken into consideration, if machine translation is to be applied to any project.


We consider and understand the challenging factors related to post-editing and offer reasonable and cost-effective solutions.