发明名称 |
System and method for speech recognition using models adapted to actual noise conditions |
摘要 |
<p>A compensation model generating unit generates compensation model to increase signal-to-noise ratio, in accordance with sequence of feature vectors produced from noise superimposed speech, upon producing the adaptive model by adapting the initial synthesized model using the utterance environmental noise model and the initial noise model. An independent claim is also included for speech recognition method.</p> |
申请公布号 |
EP1471500(A1) |
申请公布日期 |
2004.10.27 |
申请号 |
EP20040009694 |
申请日期 |
2004.04.23 |
申请人 |
PIONEER CORPORATION |
发明人 |
SEO, HIROSHI;TOYAMA, SOICHI |
分类号 |
G10L15/06;G10L15/20;(IPC1-7):G10L15/20 |
主分类号 |
G10L15/06 |
代理机构 |
|
代理人 |
|
主权项 |
|
地址 |
|