G - Physics – 10 – L
Patent
G - Physics
10
L
G10L 15/14 (2006.01) G10L 15/06 (2006.01) G10L 9/18 (1990.01)
Patent
CA 2089903
A new training method and apparatus are described for pattern-based speech recognition, e.g., a hidden Markov model (HMM) based speech recognizer using Viterbi scoring. The principle of minimum recognition error rate is applied by the present invention, in which segmentation and discriminative training are jointly optimized. Various issues related to the special structure of HMMs are presented. Parameter update expressions for HMMs are provided.
La présente invention fait état d'une nouvelle méthode de formation et d'un nouvel appareil en matière de reconnaissance de la parole basée sur les formes sonores, tel un reconnaisseur de parole basé sur un modèle de Markov caché (MMC) et utilisant une notation de Viterbi. La présente invention applique le principe du taux d'erreur de reconnaissance minimal selon lequel la segmentation et l'entraînement discriminatif sont conjointement optimisés. L'invention fait également état de diverses questions reliées à la structure particulière des MMC et fournit des expressions pour la mise à jour des paramètres des MMC.
Chou Wu
Juang Biing-Hwang
American Telephone And Telegraph Company
Kirby Eades Gale Baker
LandOfFree
Training method and apparatus for speech recognition does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Training method and apparatus for speech recognition, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Training method and apparatus for speech recognition will most certainly appreciate the feedback.
Profile ID: LFCA-PAI-O-1385603