摘要 |
PURPOSE: A continuous voice recognition device having a word dictionary of a tree structure is provided to decrease storage capacity, and to increase the recognition speed by applying a language model of a unit of a part of speech at a starting part of a word through a dictionary having a different tree structure according to the part of speech of the word. CONSTITUTION: A word cannot be discriminated with only a Korean alphabet similarly featured inverse L at a root of a tree, and the word can discriminated through a start of the tree to an end of the tree. Because a tree is sorted according to a kind of a part of speech, a part of speech pertinent to a word is grasped at a starting point. For this reason, a language model of the part of speech is able to be applied. In the conclusion, after removing a sequence of the part of speech unsuitable for a grammar, the word sequence is selected. For this reason, a recognition speed is increased. |