Continuous Speech Recognition System Using Probability Dependent Method as Backward Language Model and Its Method

摘要

1. TECHNICAL FIELD OF THE INVENTION;The present invention relates to a continuous speech recognition system and a method thereof.;2. The technical problem to be solved by the invention;The present invention is a continuous speech recognition system and method for determining the ranking of sentence recognition candidates by applying the probability dependent sentence method as a backward language model, and using the sentence structure probability calculated by the probability of transition between each word and the probability between modifier-word expression and its method. The aim is to provide a computer readable recording medium having recorded thereon a program for realizing the method.;3. Summary of Solution to Invention;The present invention comprises: feature extraction means for extracting a feature of an input speech; Word recognition means for recognizing speech on a word-by-word basis using extracted feature of input speech; Sentence recognizing means for recognizing word-wise recognition words using a forward-looking language model to recognize words in a grid structure in which the transition probability between each word is recorded; And a sentence recognizing means for recognizing sentences by backward-searching words in a lattice structure using a probability-dependent sentence method in which a probability between a modifier and a canonical word is recorded as a backward language model.;4. Important uses of the invention;The present invention is used in a speech recognition system.