The technology developed by the NRC relies on access to a parallel body of text. In this case, that would require us to have French and English equivalents at the outset in order to train the machine. That is the limitation of the current technology; if that body of text is not available, it is impossible to start up the system. In an individual case, there is always a way to look at the content, to start from somewhere, and then probably to close the loop so that the system can learn from corrections.
I would answer yes to your question. However, I would perhaps temper your expectations by telling you that it may not become a reality by next week, although the fact remains that it is possible.