for voice recognition on the device. Right now this just queries the package manager at boot and finds the (hopefully) single available recognizer. TODO: Add an attribute to let recognition services expose a settings activity, and expose the settings activity of the chosen recognition service in the system settings for voice input & output.