#artificial intelligence
Jun 2026
TF-MoE: Time-Frequency Mixture-of-Experts for Efficient Speech Separation
TF-MoE, a sparse Mixture-of-Experts (MoE) framework that enhances model capacity with almost no increase in inference cost, is proposed and demonstrated that TF-MoE consistently improves separation performance under computation cost constraints.
Qin-Zhe Hu, Chenda Li, Wangyou Zhang et al.
· arXiv.org · 3 citations
· ⚡1