Speech enhancement in realistic scenarios still remains many challenges, such as complex
background signals and data limitations. In this paper, we present a co-attention based …