Text_Password enhancement?

For my own purposes I modified Text_Password to output passwords that conform to Japanese phonology and are memorable and pronounceable by native speakers of Japanese. I'm cleaning up and extending what I did to make it produce even more natural- sounding fake Japanese and allow for options on the romanization system used. I wonder if something like this would be appropriate for submission to the general PEAR package, or if it is too specialized? It would add a couple input parameters to the end, and it is considerably longer than the other functions (it's complexity is similar to a basic search engine stemming algorithm).