Forensic possibilities in protecting victims of voice and speech cloning using artificial intelligence

The rapid advancement of modern technology has facilitated the commission

of various types of unlawful acts. The revolution in the field of

artificial intelligence (AI) has made this technology widely accessible,

significantly increasing the risk of its misuse. Voice cloning, as one of

the capabilities of AI, poses a serious threat to both individuals and

organisations. By imitating the biometric characteristics of another

person?s voice, it becomes possible to disseminate false information,

deceive security systems, and even commit criminal offences under someone

else?s identity. The growing capabilities of artificial intelligence

underscore the need to verify the authenticity of audio recordings before

conducting forensic analysis. Therefore, this paper aims to highlight the

possibilities of detecting fake audio recordings created using artificial

intelligence through a review of the available literature. Existing research

suggests that it is possible to reliably determine the authenticity of a

speaker?s voice using the auditory-spectrographic method, which remains

the predominant technique in voice and speech forensics. This approach

enables the prevention or timely detection of voice and speech cloning

attempts, thereby protecting potential victims from the potentially severe

consequences of undetected falsification. It is essential that voice and

speech experts are guided by these findings in their efforts to ensure

the reliability of the evidence entrusted to them for expert analysis, and

consequently, the outcome of the investigation.