Abstract: Recent advancements in packet loss concealment (PLC) have introduced diffusion-based generative models that offer high-quality audio reconstruction. However, their high computational costs ...