multimodal
fact
neutral
Speech deepfake detection systems can improve generalization through training recipe enhancements (multi-source data, attack-balanced sampling, audio augmentation) rather than additional architectural complexity
Artificial Intelligence02 Aug 2026