I recently released KaldiIOS framework, an iOS (Objective C) framework that provides automated speech recognition functionality on device (no need to talk to the cloud servers).

More details at:

http://keenresearch.com/kaldi-ios-framework

and a proof of concept demo app that uses the framework at:

http://github.com/keenresearch/kaldi-ios-poc

This is an early release; a number of new features, especially around making it easier to specify what the recognizer is listening for, are coming soon.

Thanks, Ogi

PS. under the hood the framework uses Kaldi (http://kaldi-asr.org), state of the art speech recognition system developed at Johns Hopkins University