Daily Papers — 2026-09-30

2 papers on audio, speech, music, and acoustics.

1. OmniReasoning: Pushing the Limits of Audio-Visual Joint Reasoning

Authors: Junming Lin, Yuxuan Wang, Zhenxin Lei, Yuxin Liu, Ruixun Liu et al.

Categories: cs.CV, cs.AI Score: 5.32/10 (Obj:7 Id:5 Ind:4 Comp:6 Eff:6 Nov:5 Rep:4)

• Strength: 在论文设置中，以 Qwen3-Omni-30B-A3B 为底座的 OmniVideoBench 准确率从 $37.2$ 升到 $50.0$（$+12.8$），且在全部音频类型与时长分组上一致为正。
• Weakness: 核心方法 MFSD 对 RLSD 在自家基准上只多 $3.3$ 分，单次运行、无方差、无组件消融，且未与 OmniVideo-R1、OmniVideo-100K、Video-KTR 等同任务方法比较。
• Full review: Claude Code 全文七公理审稿

Recent advances have enabled unified omni-modal models in understanding audio, vision, and language. However, existing benchmarks, training data, and learning methods largely treat the modalities independently, leaving the capability of audio-visual joint reasoning poorly evaluated and insufficiently elicited. We address this gap with a benchmark, data engine, and learning method. First, we introduce OmniReasoningBench, a benchmark where both audio and visual evidence are indispensable. It comprises 1,150 multiple-choice and open-ended questions across two tasks, reasoning over video and reasoning beyond video. Second, we develop a data engine OmniQA. It automatically constructs evidence-grounded QA pairs that explicitly necessitate audio-visual joint reasoning, together with time-stamped clue chains that guide the annotation of thinking process. Besides our benchmark, this engine produces training data OmniReasoning-SFT-112K and OmniReasoning-RL-19K. Finally, we propose an on-policy self-distillation method Modality-Factored Self-Distillation (MFSD). It evaluates each sampled response under modality-specific clue contexts, disentangling the contributions of individual clues and their cross-modal interactions for token-level credit assignment. With our training data and learning method, our model OmniReasoning-30B-A3B achieves 50.0% on OmniVideoBench and 42.5% on OmniReasoningBench, improving the base model Qwen3-Omni-30B-A3B-Thinking by 12.8 and 9.3 percentage points, respectively. Moreover, it delivers substantial gains on general and long-video benchmarks, including Video-MME-v2. We hope our work offers a solid step for facilitating future research in omni-modal joint reasoning.

2. UniAE-MoE: A Unified Audio Encoder via Mixture of Experts

Authors: Shengbo Cai, Zhisheng Zhang, Zichao Nie, Jing Peng, Jingran Xie et al.

Categories: cs.SD, cs.AI Score: 4.42/10 (Obj:6 Id:4 Ind:3 Comp:4 Eff:5 Nov:3 Rep:7)

• Strength: 双 encoder + MoE 融合在 Interspeech 2026 Audio Encoder Challenge Track A 拿到 23 队最高总分 $0.9122$，代码、权重、逐条预测全部开源。
• Weakness: 相对 Concat 的 $+0.024$ 有 $76\%$ 集中在三个任务，FMA 的 $0.848\to0.979$ 很可能来自 fmamedium 训练集对 fmasmall 测试集的泄漏，且没有参数匹配的 dense 基线和多 seed 方差。
• Full review: Claude Code 全文七公理审稿

Large Audio Language Models (LALMs) rely on effective audio encoders for multi-task performance. We introduce UniAE-MoE, a unified audio encoder designed to model cross-domain audio representations and achieve outstanding downstream understanding performance via a Mixture-of-Experts (MoE) architecture. Specifically, we explore mainstream audio encoders and integrate those from Qwen2-Audio and Audio-Flamingo 3, which demonstrate superior downstream capabilities. To facilitate effective model fusion, we improve our encoder using SwiGLU with shared experts to decouple encoder networks, and we further introduce a two-stage instruction-tuning strategy to better adapt the model to diverse downstream tasks. Moreover, we propose the task-specific data scaling (TSDS) technique to enhance \tool’s understanding capabilities. On the XARES-LLM benchmark, UniAE-MoE attains a score of 0.802, achieving state-of-the-art performance. It also delivers top-tier performance in the official Interspeech 2026 Audio Encoder Capability Challenge, further demonstrating robust generalization across diverse audio tasks. Together, these results validate the effectiveness of \tool for unified audio understanding across speech, music, and general audio domains.