arXivMing-Hao Hsu, Shujie Liu, Yuxuan Hu, Yuxiang Wang, Yingda Shen, Jinyu Li, Zhizheng WuTue, Nov 3, 2026, 12:12 AM PST
score 16.7
Frozen Audio Model Finds Answer Clips in Hour-Long Recordings
Original: ReListen: Training-Free Evidence Localization in Hour-Long Audio with Frozen Audio Language Models
Source: arxiv.org ↗
Writing ELI5 summary…