AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

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

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Speech and audio language models

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add SFT on EAQA-SFT to the base model (audio-grounded format) (ablation)Qwen2.5-Omni-7B · MMARaccuracy: 51.80 -> 56.77Improved
switch inference of the SFT cold-start model to audio-interleaved format without RL (ablation)Cold-start model · MMARaccuracy: 56.77 -> 52.26Hurt
add RL after interleaved adaptation (ablation)Echo · MMARaccuracy: 52.26 -> 69.99Improved
use audio-interleaved format instead of audio-grounded text format under the same RL (ablation)Echo · MMARaccuracy: 64.63 -> 69.99Improved
add SFT before RL (audio-grounded) vs direct RL on base format (ablation)Qwen2.5-Omni-7B · MMARaccuracy: 63.15 -> 64.63Improved
audio-interleaved reasoning adds inference cost (ablation)Echo · MMARreasoning latency: 2.06 s -> 2.12 s (direct RL vs Echo)No clear effect
use EAQA-RL (21.9k) instead of AVQA (40.4k) as RL data (ablation)Echo · MMARaccuracy: 67.58 -> 69.99Improved
use EAQA-RL instead of AVQA for direct RL (ablation)Direct RL model · MMARaccuracy: 59.51 -> 63.15Improved
filter SFT data with re-evaluation (75,862 high quality vs 111,868 unfiltered) (ablation)Echo · MMARaccuracy: 64.67 -> 69.99Improved
filter RL data with re-evaluation (ablation)Echo · MMARaccuracy: 67.03 -> 69.99Improved
use stronger audio extractor Qwen2.5-Omni instead of Qwen2-Audio for data generation (ablation)Echo data pipeline · MMARaccuracy: 63.91 -> 69.99Improved
use Qwen2.5-Omni instead of Kimi-Audio as audio extractor (ablation)Echo data pipeline · MMARaccuracy: 68.82 -> 69.99Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy