| 摘要 |
A speech recognition device which can preferably be used for reducing the memory capacity required for speaker independent speech recognition is provided. A matching unit 114 spreads speech models belonging to a first speech model network 300 and a garbage model 350 in a RAM 112, and gives a speech parameter extracted by a speech parameter extraction unit 106 to the speech model in the RAM 112, and when an occurrence probability output from the garbage model 350 is equal to or greater than a predetermined value, the matching unit 114 spreads speech models belonging to any of speech model groups 402 in the RAM 112 based on the occurrence probability output from the speech model belonging to the first speech model network 300.
|
