SPELL MODE IN A SPEECH RECOGNIZER

摘要

A speech recognition system includes an input for receiving a signal representative of speech (SIS) and a recognition unit for recognizing the signal and representing the recognized signal as word(s). The recognition unit includes a spell mode recognizer which compares the signal against a spell inventory comprising a representation of a plurality of spell reference words. The spell mode recognizer identifies a single spell character associated with a most likely one of the spell reference words. The spell inventory includes at least two multi-character spell reference words associated with a same single spell character.