Voiceless fricative sounds and forensic implications

Abstract

This paper presents the results of a study aimed at assessing the capability of Spanish unvoiced fricative sounds to discriminate between speakers in forensic phonetics. The center of gravity, standard deviation, skewness, kurtosis and higher peak intensity of the spectra of fricative segments, as well as the standardized values of the bands LTAS have been analyzed. Results achieved on average 89.78% of original grouped cases correctly classified, and 71.6% in cross-validation. Theoretical implications are discussed, as well as its applicability in forensic speaker identification with a closed reference set.