08-10-2017, 05:41 PM
Given that these passwords are generated in a semi-random way, with rules around how they are produced to attempt to make them "pronounceable" it may be possible to reduce the keyspace a bit or at least generate markov data based on a large set of these "pronounceable" passwords to increase chances of finding it slightly. That said, still a huge space to be searching, even with any sort of optimizations. If you know any parts or can guess as to what they may have been, you can reduce the keyspace further and try to bring it into feasibility, but its going to be a lot of guessing and luck.