TY  - RPRT
TI  - Boosting ASR Robustness via Test-Time Reinforcement Learning with Audio-Text Semantic Rewards
AU  - Linghan Fang
AU  - Tianxin Xie
AU  - Li Liu
PY  - 2026
UR  - https://arxiv.org/abs/2603.05231
ID  - 2603.05231
ER  -