Abstract: A combination of a series of filters and a specification language is used to rapidly enable voice access to an external data repository. The filters include a data-to-voice filter operating on data values that flow from the data repository to a communication system, a voice-to-data filter capturing voice inputs and storing data related to such inputs in the data repository, an utterance filter normalizing spoken data inputs in a particular format, a validation filter checking values returned by the speech recognizer, and a data description filter creating spoken formats of data labels or descriptions. Also included is a pair of dictionaries: name pronunciation dictionary for ensuring correct pronunciation of words and a name grammar dictionary providing variations associated with a name.