TY  - RPRT
TI  - What Makes Codec Tokens Effective for Target Speaker Extraction? A Semantic-Acoustic Disentanglement Perspective
AU  - Wang Xiang
AU  - Bang Zeng
AU  - Ming Li
PY  - 2026
UR  - https://arxiv.org/abs/2610.16858
ID  - 2610.16858
ER  -