Method of training neural networks used for speech recognition

Details Method of training neural networks used for speech recognition Method of training neural networks used for speech recognition

G - Physics

06

F

G06F 15/18 (2006.01) G06N 3/04 (2006.01) G10L 15/16 (2006.01) G10L 15/26 (2006.01) G10L 15/02 (2006.01) G10L 15/06 (2006.01) G10L 15/10 (2006.01) G10L 15/22 (2006.01)

Patent

CA 2190631

A speech-recognition system for recognizing isolated words includes pre-processing circuitry (3) for performing analog-to-digital conversion and cepstral analysis, and a plurality of neural networks (12-14) which compute discriminant functions based on polynomial expansions. The system may be implemented using either hardware or software or a combination thereof. The speech waveform of a spoken word is analyzed and converted into a sequence of data frames. The sequence of frames is partitioned into data blocks, and the data blocks are then broadcast to a plurality of neural networks. Using the data blocks, the neural networks compute polynomial expansions. The output of the neural networks is used to determine the identity of the spoken word. The neural networks utilize a training algorithm which does not require repetitive training and which yields a global minimum to each given set of training examples.

Un système de reconnaissance de la parole permettant de reconnaître des mots isolés comprend un circuit (3) de prétraitement qui effectue la conversion analogique-numérique et l'analyse cepstrale et une pluralité de réseaux neuronaux (12)-(14) qui calculent des fonctions discriminantes fondées sur des expansions polynomiales. Ce système peut être mis en oeuvre avec des machines et/ou des logiciels. Le signal vocal d'un mot énoncé est analysé et converti en une séquence de trames de données. Cette séquence de trames est découpée en blocs de données qui sont ensuite envoyés à une pluralité de réseaux neuronaux. A l'aide de ces blocs de données les réseaux neuronaux calculent des expansions polynomiales. La sortie desdits réseaux neuronaux est utilisée pour déterminer l'identité du mot énoncé. Les réseaux neuronaux utilisent un algorithme de formation qui ne nécessite pas d'apprentissage répétitif et qui produit un minimum global pour chaque ensemble donné d'exemples d'apprentissage.

Affiliated with

Also associated with

Rating

Method of training neural networks used 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 Method of training neural networks used for speech recognition, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of training neural networks used for speech recognition will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFCA-PAI-O-1706441