Echo: Towards Advanced Audio Comprehension via Audio-Interleaved Reasoning
Daiqing Wu, Xuan Zhang, Dongbao Yang, Jiashu Yao, Longfei Chen, Qingsong Liu, Sicheng Zhao, Can Ma, Yangyang Kang, Yu Zhou · 2026
Read Echo: Towards Advanced Audio Comprehension via Audio-Interleaved Reasoning by Daiqing Wu, Xuan Zhang, Dongbao Yang, Jiashu Yao, Longfei Chen, Qingsong Liu, Sicheng Zhao, Can Ma, Yangyang Kang, Yu Zhou (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-11909.txt.
Description
Large Audio Language Models are expected to comprehend complex audio much like humans, but current efforts mostly replicate text-based reasoning by encoding audio once, creating an information bottleneck. Daiqing Wu and nine co-authors propose audio-interleaved thought, treating audio as an active thought component for sustained engagement and perception-grounded analysis. A two-stage training framework first teaches models to find salient audio segments through supervised fine-tuning, then rewards skilled re-listening through reinforcement learning.
During thought, the model can re-listen on demand and refine its grasp through a structured generation pipeline, moving beyond one-time coding. On audio comprehension benchmarks, audio-woven thought leads overall in both efficiency and broad use. The authors present the approach as a strong path for advancing audio grasp, arguing that sustained listening, not single-pass encoding, is what audio understanding requires.
Catalog record
| CREATOR | Daiqing Wu, Xuan Zhang, Dongbao Yang, Jiashu Yao, Longfei Chen, Qingsong Liu, Sicheng Zhao, Can Ma, Yangyang Kang, Yu Zhou |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 82 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-11909