Assessing the discriminative power of voice / Pasupathy Naresh Trilok
Idioma: Inglês.Publicação: Westchester County, NY : Pace University, 2004Descrição física: 1 CD-ROM; 12 cmAssunto: 346274 | 345486 | 334284 | 334103Resumo: "This study establishes the individuality of voice, using the discriminative power of biometric data quantitatively. The task of establishing voice modality to discriminate every person is difficult because there are a large number of classes for the entire population. The paper proposes a methodology that is statistically inferable. A many class problem is transformed into a dichotomy by using distances between measurements of intra and inter-person classes. This establishes a thorough distinction of classes and thereby validates distinct individuality. This model remains statistically inferable even when it does not observe all the classes."| Item type | Current location | Call number | Copy number | Status | Date due | Barcode |
|---|---|---|---|---|---|---|
| Monografias | Biblioteca da Polícia Judiciária | PJCD143 (Browse shelf) | PJCD143 | Available | PJ1CD143_47 |
"Master’s Dissertation - School of Computer Science and Information System - Pace University - Supervised by Dr. Sung-Hyuk Cha, Dr. Charles Tappert".Disponível em: https://support.csis.pace.edu/CSISWeb/docs/MSThesis/TrilokPasupathyNaresh.pdf (acedido em 20 de Junho de 2011). Ficheiro de 1,03 Mb em formato pdf (57 p.).
"This study establishes the individuality of voice, using the discriminative power of biometric data quantitatively. The task of establishing voice modality to discriminate every person is difficult because there are a large number of classes for the entire population. The paper proposes a methodology that is statistically inferable. A many class problem is transformed into a dichotomy by using distances between measurements of intra and inter-person classes. This establishes a thorough distinction of classes and thereby validates distinct individuality. This model remains statistically inferable even when it does not observe all the classes."
There are no comments for this item.