Current releases of CopiaFacts incorporate features which allow interfacing to various speech recognition packages.  You can use scripting to pass recorded voice files to third-party recognition engines, and if you record IVR responses into a memory buffer (see RECORD_BUFFER), you can pass these to a custom DLL which calls the API of a recognition engine.

Copia do not provide interfaces to any current speech-recognition engines, but can advise in more detail on request about how to implement this feature in an IVR application.