DIGITAL WATERMARKING FOR PROTECTING AUDIO CLASSIFICATION DATASETS

In this paper, we investigate plausibility about encoding an example into extent about transient recurrence portrayal about a subset about sound order datasets utilized in profound learning. According towards previous research, actual sound about watermarked audio must be used towards extract information contained within it. An audio watermarking system that only uses classification results towards determine whether or not a deep learning-based audio classification model was trained on watermarked audio classification dataset is subject about this article. Our suggested method is capable about determining use about an audio classification dataset, as demonstrated by experimental results, among little effect on overall classification performance.