Speech recognition, CMU Sphinx

Wondering if anyone has any experience with this cross-platform speech recognition library Sphinx from Carnegie Mellon.
I am looking to try adding speech commands in my game and it needs to be cross-platform, so the Windows speech SDK (which has third-party Python wrappers) is not going to cut it.

i did work with sphinx2 i think. long ago. was quite a rough experience back then, setting up, training,…

found simon to be a lot easier to set up and work with.

worked pretty well for me. altho i cant stand listening to my own voice, you can find an example here.
used regular keyboard bindings instead of direct python integration.

as you can see, it does a pretty good job even without extensive training (i went with the defaults)

Anyone out there interested in integrating Sphinx with Panda3D if there was a financial bounty for doing so? This might make a nice CMU student project. :wink:
Ideally it would be fully integrated (accessible with Python), able to run in a thread, and be able to send the recognized words using Panda’s messenger/accept system.
Of course the whole community would also benefit.