DECIPHERING AUDIO DEEPFAKES: AN INVESTIGATION BY FREQUENCY ANALYSIS

Abstract

This research examines the emerging technology of audio DeepFakes, which presents itself as a contemporary challenge. To address this issue, ElevenLabs was used for voice cloning, along with scripts based on Deep Learning and Machine Learning for the creation of spectrograms and evaluation of patterns in the frequencies of the voices. The findings indicated that the evaluation of the graphs of each voice is efficient to differentiate the artificial voice from the real one.