Integrating Spectro-Temporal Cross Aggregation and Multi-Scale Dynamic Learning for Audio Deepfake Detection | IEEE Conference Publication | IEEE Xplore