Audio Dub Detection Using Machine Learning

The advancements in the field of audio editing tools has significantly increased the prevalence of manipulated audio recordings, posing serious challenges across various fields such as journalism, legal proceedings, cyber security, and personal privacy. These tools enable the creation of highly convincing fake audio, which complicates the task of distinguishing genuine recordings from altered ones. Current detection methods often rely on complex and resource-intensive algorithms, which are not only difficult to implement but also require significant computational resources. This study proposes a straightforward approach using simple machine-learning algorithms to detect fake audio. By applying various pre-processing techniques to remove outliers and enhance accuracy, the Support Vector Machine (SVM) method achieves an optimized accuracy rate of 98.785%.