METHOD AND SYSTEM FOR MODELING COMMON-LANGUAGE SPEECH RECOGNITION IN COMPUTER WITH BACKGROUND OF A PLURALITY OF DIALECTS

PROBLEM TO BE SOLVED: To provide a system for increasing the recognition rate of a dialect accent common-language while data quantity is small, and simultaneously prevent extreme decrease in recognition rate of a standard common-language at the same time. SOLUTION: This method includes building the standard triphone-based common-language model based on the training data of the standard common-language, and building first and second monophone-based dialect accent common-language models respectively based on the development data of the first and second dialect accent common-languages. This method further includes acquiring a first confusion matrix through recognizing the development data of the first dialect accent common-language using the standard common-language model, and merging the first dialect accent common-language into the standard common-language model to acquire the temporary merging model according to the first confusion matrix. This method further includes acquiring a second confusion matrix through recognizing the development data of the second dialect accent common-language according to the temporary merging model, and merging the second dialect accent common-language model into the temporary merging model to acquire the recognition model. COPYRIGHT: (C)2010,JPO&INPIT