Deepfake Audio
Deepfake audio, the artificial creation or manipulation of audio recordings using AI, poses a significant threat to authenticity and security. Current research focuses on developing robust detection methods, employing various neural network architectures like convolutional neural networks (CNNs) and recurrent neural networks (RNNs), often incorporating multimodal analysis combining audio and visual cues, or leveraging self-supervised learning and continual learning to adapt to evolving deepfake techniques. This field is crucial due to the potential for widespread misuse in misinformation campaigns, fraud, and identity theft, driving the need for effective detection and attribution systems.
Papers
Codecfake: An Initial Dataset for Detecting LLM-based Deepfake Audio
Yi Lu, Yuankun Xie, Ruibo Fu, Zhengqi Wen, Jianhua Tao, Zhiyong Wang, Xin Qi, Xuefei Liu, Yongwei Li, Yukun Liu, Xiaopeng Wang, Shuchen Shi
FakeSound: Deepfake General Audio Detection
Zeyu Xie, Baihan Li, Xuenan Xu, Zheng Liang, Kai Yu, Mengyue Wu