| 摘要 |
PROBLEM TO BE SOLVED: To provide a speech understanding apparatus which accurately understands an utterance. SOLUTION: The speech understanding apparatus 1 has: a speech recognition unit 20 which performs speech recognition of the utterance using multiple different language models; a language understanding unit 30 which uses the multiple different language understanding models to perform language understanding for each of speech recognition results output by the speech recognition unit 20 using the multiple different language model, and obtains multiple speech understanding results; and an integrating unit 40 which, on the basis of each of the speech understanding results, calculates utterance batch confidences, based on features obtained from likelihood of a sound model, the number of included concepts and confidence of each concept, and selects and outputs one of the speech understanding results with a highest utterance batch confidence among the calculated utterance batch confidences. COPYRIGHT: (C)2010,JPO&INPIT |
