1 paper · 1 filter
Dohwan Kim, Jung-Woo Choi
In speech enhancement, knowledge distillation (KD) compresses models by transferring a high-capacity teacher's knowledge to a compact student. However, conventional KD methods trai…