AudioVeritas: A Machine Learning Model to Detect Deepfake Audio
M Ganavi · International Journal for Research in Applied Science and Engineering Technology · 2025
With the rapid advancement of deep learning technologies, the creation of synthetic media, particularly deepfake audio, has become increasingly prevalent. Deepfake audio convincingly replicates human voices, presenting both innovative opportunities and significant risks, including misinformation and fraud. Detecting such audio is challenging due to the subtle differences from real speech, often imperceptible to human hearing. This paper introduces a machine learning-based framework for detecting and classifying audio as "REAL" or "DEEPFAKE." Leveraging signal processing techniques, including MelFrequency Cepstral Coefficients (MFCCs), Chroma Features, and Spectral Properties, our approach identifies patterns distinguishing real audio from synthetic. The proposed system addresses ethical and security concerns surrounding synthetic audio misuse while contributing to advancements in deepfake detection research.