The project involves the development of a new automatic method of parallel corpora alignment. This new approach will be based on Neural Machine Translation and previous aligned corpora. The method will be tested on a Czech-English parallel corpus of Faculty news, which alignment will be improved as a result.