DYNAMIC PHONEME DICTIONARY FOR SPEECH RECOGNITION

A phoneme dictionary for speech recognition has a continuously stored set of standard words and their phonemes. As documents are needed by an application using the phoneme dictionary, words in the documents are selected and dynamically and temporarily added to the phoneme dictionary. Phonemes corresponding to the added words are generated using a text-to-phoneme converter, and are then temporarily added to the dictionary. Temporarily added words and their phonemes are dropped from the phoneme dictionary when their document is no longer needed by the application.