We report here the results of a first timid attempt to promote an evaluation campaign on a Forensic Speaker Identity Verification task within Evalita 2009. Participants were prompted to test methods and models usually used in forensics on a common corpus collected simulating real forensic characteristics and situations. The Task presented a Training data set including known suspected voices to be compared with voices in two other data sets, namely a Closed-test set of 16 unknown voices and an Open-test set containing different voices to be segmented before the test or comparison. Results achieved by participants are here briefly reported.
Forensic Speaker Identity Verification (F-SIV) in Italy First Evaluation Campaign Evalita-2009
ROMITO, Luciano;
2009-01-01
Abstract
We report here the results of a first timid attempt to promote an evaluation campaign on a Forensic Speaker Identity Verification task within Evalita 2009. Participants were prompted to test methods and models usually used in forensics on a common corpus collected simulating real forensic characteristics and situations. The Task presented a Training data set including known suspected voices to be compared with voices in two other data sets, namely a Closed-test set of 16 unknown voices and an Open-test set containing different voices to be segmented before the test or comparison. Results achieved by participants are here briefly reported.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
