5th INTERNATIONAL CONGRESS ON KHAZAR SCIENTIFIC RESEARCH AND INNOVATION
DEEPFAKE AND FORENSIC AUDIO ANALYSIS: NEW GENERATİON DIGITAL THREATS AND METHODS FOR DISTINGUISHING DEEPFAKE VOICES
The integration of technology into our lives has simplified our daily routines in many ways, yet it has also introduced various risks. With the continuous renewal and advancement of the digital world, technology has started to branch out into diverse areas. One of these areas is Deepfake, a term we first started hearing about in 2017, known in Turkish as "Derin Sahtecilik." Deepfake technology is a form of artificial intelligence that uses machine learning algorithms to create highly realistic, yet fabricated, content such as synthetic audio and visuals, making it seem as though nonexistent recordings are real. The synthetic voices generated by artificial intelligence closely mimic real voices, and distinguishing them from genuine voices has become increasingly difficult. Deepfake audio poses serious challenges. For example, issues like identity theft and fraud can significantly impact individuals' social and economic lives. In legal investigations, deepfake content can mislead by introducing fabricated evidence, potentially leading to wrongful convictions of individuals or institutions. This situation heightens the importance of forensic audio analysis methods used to verify the authenticity and source of audio; however, as deepfake audio continues to become more realistic, these analyses are increasingly challenging. This study examines the new threats that Deepfake technology poses to forensic audio analysis and discusses the differentiation methods available to counter these threats.
