Shallow Diffusion For Fast Speech Enhancement Papers