| 摘要 |
PURPOSE:To expand the content of a dictionary for word recognition and to efficiently recognize voices of ultra-large vocabularies with a high recognition rate by recognizing input voices as words or clauses in accordance with the characteristic parameters of vocal sound units, displaying the results of the recognition, deciding one recognition result as the recognition result and registrating the recognition result in the dictionary. CONSTITUTION:The characteristic parameters of input voice signals are recognized by the vocal sound units in a word and clause recognizing section 6 and are respectively referred to dictionaries 9, 11 for clause and word recognition. The recognition by clause and word recognizing section 8, 10 is then executed. The recognition results are displayed on a display section 12 and the optimum decided word of the one recognition is registered via a keyboard 13 in the dictionary 11. This device is characterized by having a dictionary register means. The word recognition dictionary is expanded in this way and the voices of the ultra-large vocabularies with the vocal sounds as the recognition units are efficiently recognized at the high recognition rate. |
