Speaker verification system using decision fusion logic
US5839103A · kind A · utility
97Cited by
10References
24Claims
0Family size
Assignee
Inventors
Key dates
| Filing date | Jun 7, 1995 |
| Grant date | Nov 17, 1998 |
| Priority date | — |
| Expiry date | Jun 7, 2015 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG10L25/03
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
The present invention relates to a pattern recognition system which uses data fusion to combine data from a plurality of extracted features and a plurality of classifiers. Speaker patterns can be accurately verified with the combination of discriminant based and distortion based classifiers. A novel approach using a training set of a "leave one out" data can be used for training the system with a reduced data set. Extracted features can be improved with a pole filtered method for reducing channel effects and an affine transformation for improving the correlation between training and testing data.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.