Looking at the app's video, it appears to be a directed soundboard with a tree structure, parts of speech, and queueing of words:

http://www.youtube.com/watch?v=VpafqM1kiNo

Even if the app got sued out of existence, implementing the same functionality would be fairly trivial.

I'm assuming that the unique, patentable portions are a lot of "when used in this way" clauses, and the speech research to come up with the contents of the soundboards.

The end run around this would be an generic soundboard app with a bunch of silly noises but with the ability to load other soundboards and rulesets in a set data format that would implement the functionality.