 |
ITRW on Non-Linear Speech Processing (NOLISP 05)Barcelona, Spain |
 |
|
ITRW on Non-Linear Speech Processing (NOLISP 05)Barcelona, Spain |
|
Threshold estimation methods mainly deal with the scarcity of data and the difficulty of obtaining data from impostors in real applications. In this context, potential outliers, i.e., those client scores which are distant with respect to mean, could lead to wrong mean and variance client estimations, which are commonly used by somel threshold estimation methods. To alleviate this problem, some efficient threshold estimation methods based on weighting scores are proposed here. Before estimating the threshold, the set of client scores is totally or partially weighted, improving subsequent estimations. The weighting factor is obtained from a non-linear function that distributes scores according to their distance to the estimated mean. Text-dependent experiments have been carried out by using a telephonic multi-session database in Spanish. The database has been recorded by the authors and has 184 speakers.
Bibliographic reference. Saeta, Javier R. / Hernando, Javier (2005): "New speaker-dependent threshold estimation method in speaker verification based on weighting scores", In NOLISP-2005, 34-41.