SYSTEM FOR ENTOLLEMENT OF A SPOKEN NAME DIALING SERVICE
摘要
Spoken name speed dialing is accomplished through a media server which compares a digitized audio stream to speaker dependent and speaker independent models. At least some of the speaker dependent models are generated dynamically for each voice recognition session. The speaker dependent models are penalized according to their respective lengths in order to improve out-of-vocabulary rejection while minimizing any adverse effects on in-vocabulary performance.
申请公布号
WO9923641(A1)
申请公布日期
1999.05.14
申请号
WO1998US22295
申请日期
1998.10.21
申请人
ALCATEL USA SOURCING, L.P.
发明人
FISHER, THOMAS, D.;MOWRY, DEARBORN, R.;SPIESS, JEFFREY, J.