A Multi-Domain Feature Framework for Robust Deepfake Audio Detection
The results indicate that multi-domain feature fusion offers a practical and generalizable solution for real-world deepfake audio detection, particularly in environments involving diverse codecs and synthesis techniques.