Speaker identification and verification using Gaussian mixture speaker models
Speech Communication
The inference of identity in forensic speaker recognition
Speech Communication - Speaker recognition and its commercial and forensic applications
AHUMADA: A large speech corpus in Spanish for speaker characterization and identification
Speech Communication - Speaker recognition and its commercial and forensic applications
Toward patient identification using chest CT scan
WBMA '03 Proceedings of the 2003 ACM SIGMM workshop on Biometrics methods and applications
Fast and Accurate 3D Face Recognition
International Journal of Computer Vision
| Hi-index | 0.00 |
On the one hand, commercial biometric systems and forensic identification require different approaches in order to evaluate system outputs. On the other hand, bayesian approach for evidence analysis and forensic reporting perfectly suits the needs of the court and the forensic scientist. Inside this bayesian framework, any biometric system can be adapted to provide its results in the form of likelihood ratios (LR) (being so converted in a forensic identification system), and performance of the forensic system can be then assessed according to the bayesian approach. We will focus on a specific biometric characteristic, showing how forensic speaker recognition can be reported by means of bayesian technique. Results including NIST-Ahumada and providing LR scores in the form of Tippet plots (and compared with DET plots) will be finally presented.