PocketSphinx seems to be a very robust, open-source speech recognition system that has the advantage of working offline... for those occasions where internet access to back-end Google servers isn't an option. www.cs.cmu.edu/~awb/papers/ICASSP2006/0100185.pdf
While it seems someone has ported the project to Android, which another thread referenced, My Droid Experience: Offline Speech Recognition With PocketSphinx, the instructions listed seem to be for those with Linux access and a good deal of familiarity with that system.
Is there a way we could have an easily importable library for projects which would implement PocketSphinx?