The last 12 articles in my "blog" (
http://web.onetel.com/~hibou/blog/Blog.html) cover my statistical analysis of the Voynich Manuscript. I have concluded that the text is almost certainly meaningless, and figured out how it could have been generated. My generated text has all the important statistical properties of the original, except for some details of paragraphing, which can easily be handled by postprocessing.
There's a related thread discussing it at Voynich Ninja: https://voynich.ninja/thread-1812.html