Forensic phonetic analysis of voice in the criminalistic identification of speakers

This paper examines the role and significance of forensic-phonetic voice analysis in the criminalistic identification of speakers, with a particular focus on its application under real investigative and judicial conditions. Voice is considered a specific criminalistic trace of intangible nature, whose identification value arises from a combination of physiological, behavioural, phonetic, and linguistic characteristics of the speaker. Unlike controlled biometric systems, forensic speaker recognition is performed in uncontrolled and often degraded conditions, which requires a cautious and evaluative approach to interpretation. The paper is based on a review and comparative analysis of relevant domestic and international literature in the fields of forensic phonetics, forensic speaker comparison, and criminalistics. Special attention is given to methodological approaches used in practice, including auditory-perceptual, acoustic-instrumental, and combined methods, as well as to the problem of intra- and inter-speaker variability and its impact on the reliability of conclusions. The paper highlights the importance of an evaluative framework in forensic voice comparison, particularly the use of the likelihood ratio approach, which allows the strength of the evidence to be assessed relative to competing hypotheses rather than through categorical identification. The growing role of automated and machine-learning-based systems is also discussed, emphasizing that such tools should be empirically validated and used as support to expert judgment, not as a substitute for it. The findings indicate that a combined expert-based and instrumental approach, aligned with international standards such as those of the European Network of Forensic Science Institutes (ENFSI), currently represents the most reliable model for forensic speaker identification. The paper concludes with recommendations aimed at improving forensic and criminalistic practice, including standardization of reference samples, transparent reporting of limitations, and continuous education of criminal justice professionals on the interpretation of forensic voice evidence.