Acoustic and auditory relation of creak/creaky settings in telephone call and direct recording samples: An approach in the light of forensic phonetics

Abstract

This study assessed the possible relationship between acoustic and auditory analyses of vocal samples extracted from telephone calls and direct recordings. Samples from a database of auditory assessments of Voice Quality (VQ) performed by experienced phoneticians using Vocal Profile Analysis were used. Analyses were made of recordings of spontaneous speech samples from five men (aged 20–56, diagnosed with dysphonia), obtained simultaneously by telephone calls and direct recordings. Acoustic analysis was conducted by extracting 20 melodic measurements and VQ descriptors, in an automated manner using a script in the PRAAT software. These measurements were then compared with the results of database analyses using descriptive statistics. The results showed that, despite the differences obtained between the recording channels, acoustic parameters associated with voice quality, such as jitter, shimmer, and harmonic-to-noise, presented consistent correlations with perceptual judgments of non-modal phonation, especially in segments with creak or creaky voice. The analysis revealed that the telephone channel tended to smooth out abrupt peaks in the acoustic measurements, but did not compromise the identification of predominant voice quality patterns. This stability suggests that, even with the inherent signal degradation in telephone transmission, automated acoustic analyses can provide relevant and reliable information for forensic phonetic analysis. We conclude that the integration of acoustic and perceptual analyses, even using recordings from different channels, is feasible and valuable for the forensic assessment of voice quality. These findings highlight the importance of a careful methodological design that considers the particularities of the recording channels and reinforce the potential of forensic phonetics to provide robust evidence in legal contexts.