Abstract

The present work analyzed the effect of sex, noise, vocal register, and channel on the performance of Batvox, the automatic speech recognizer employed by the Forensic Laboratory of the Investigative Police of Chile. This study transferred and applied the methodology and results of a previous and recently finished research project [Regular Fondecyt No. 1,070,210, funded by the National Commission for Scientific and Technological Research of the Government of Chile] to develop a model for the generation of speech reference populations with forensic purposes, which incorporated a set of dialect and environmental variables not considered in the selected samples currently integrating their databases. The information supplied provided methodological criteria to develop an optimized database for the biometric identification of speakers in Chile and to improve the performance of the systems applying them.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call