This AI Paper Proposes a Novel Dual-Branch Encoder-Decoder Architecture for Unsupervised Speech Enhancement (SE)