Utilizing multiple speech recognizers to improve spoken language understanding performance

This paper proposes the method of enhancing the spoken language understanding performance by employing general domain and domain specific speech recognizer. Apply intent to the result from general domain and domain specifically trained speech recognizer. Verify that lexico-semantic pattern matching is the target pattern, then calculate phoneme string similarity from the named entity result and select the intent. Result of the experiment indicates that the proposed method satisfies both general domain and target domain performances.