Elamur Research⌕
☰
home / papers / 2026 / Don't Just Listen, Try Planning: Graph-based Retrieval-Generation Agent for Long-form Audio Meeting Understanding

Don't Just Listen, Try Planning: Graph-based Retrieval-Generation Agent for Long-form Audio Meeting Understanding

Paper · 2026 · Quanwei Tang, Dong Zhang, Shoushan Li et al.

Summary

Question answering over long multi-party meeting recordings has had little dedicated data, and current approaches struggle with it. Speech LLMs mostly reduce audio to ASR text, which discards cues such as loudness and tone (the authors call this "acoustic missing"). They also lose track of evidence spread across tens of minutes ("context fragmentation"). Plain text or audio RAG does not fix the second problem either.

The authors make two contributions:

  • LongAudioQA, a benchmark built from AliMeeting, AMI and DailyTalk. Its questions were generated by LLMs and then checked by humans over several rounds.
  • GRGA, an agent that stores each recording as a multi-dimensional heterogeneous graph. The graph links utterances by semantics, speaker attributes, timestamps and acoustic features. The agent then plans retrieval steps, runs search tools over the graph, critiques the evidence it found and re-plans before answering. This loop is framed as a POMDP.

According to the authors, GRGA clearly beats both end-to-end Speech LLMs and RAG-based state-of-the-art systems in automatic and human evaluation. The available excerpt does not include the exact scores. The work matters because it pairs a benchmark with a working method for reasoning over hour-scale audio, including paralinguistic cues.

Key points

  • LongAudioQA statistics:
    • AliMeeting: 14.91 h, 28 dialogues, 3,013 QA pairs, about 1,935 s and 815 turns per dialogue.
    • AMI: 18.24 h, 34 dialogues, 3,243 QA pairs.
    • DailyTalk: 21.59 h, 128 dialogues, 10,200 QA pairs; each input joins 20 clips, giving under 10 minutes.
  • Question taxonomy: five categories, namely Factual, Inferential (multi-hop), Temporal, Summarization and Acoustic-Aware.
  • Quality control:
    • Timestamp evidence must reach IoU > 0.9 with the ground truth.
    • The paper reports Cohen's κ = 0.91 in the dataset section but 0.95 in the Limitations section, an internal inconsistency.
  • GRGA components:
    • Agent modules: Query Decomposer, Execution Planner and Self-Reflector.
    • Tools: keyword_search, semantic_search, time_range_search, filter_by_speaker and hybrid_search.
    • Graph edge types: temporal, ReplyTo, SameSpeaker, semantic/reference and entity edges.
  • Stated limitations:
    • Errors from upstream ASR and diarization propagate into the graph.
    • Iterative planning makes inference slower than single-pass RAG.
    • Evaluation covers only structured meetings.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Soochow UniversitySuzhou · China63 works · 4,507 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSignal Processing · 452 works

🧭 Topics 2Research subjects and languages the work is about.Information retrieval & search · 473 worksSpeech & spoken language · 480 works

Cite

Cite as Quanwei Tang, Dong Zhang, Shoushan Li and Guodong Zhou. 2026. Don't Just Listen, Try Planning: Graph-based Retrieval-Generation Agent for Long-form Audio Meeting Understanding. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1038

@inproceedings{tang2026don,
  title = {{Don't Just Listen, Try Planning: Graph-based Retrieval-Generation Agent for Long-form Audio Meeting Understanding}},
  author = {Quanwei Tang, Dong Zhang, Shoushan Li and Guodong Zhou},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1038},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1038}
}

References

38 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mohammad Mahdi Abootorabi and Ehsaneddin Asgari. 2025. Clasp: Contrastive language-speech pretraining for multilingual multimodal information retrieval. In Advances in Information Retrieval: 47th European Conference on Information Retrieval, ECIR 2025, Lucca, Italy, April 6–10, 2025, Proceedings, Pa… find
  2. Jianlv Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2024a. Bge m3-embedding: Multi-lingual, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. Preprint, arXiv:2402.03216.
  3. Yafeng Chen, Siqi Zheng, Hui Wang, Luyao Cheng, Qian Chen, Shiliang Zhang, and Junjie Li. 2024b. Eres2netv2: Boosting short-duration speaker verification performance with computational efficiency. Preprint, arXiv:2406.02167. — [arXiv:2406.02167](https://arxiv.org/abs/2406.02167) find
  4. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, Luke Marris, Sam Petulla, Colin Gaffney, Asaf Aharoni, Nathan Lintz, Tiago Cardal Pais, Henrik Jacobsson, Idan Szpektor, Nan-Jiang Jiang, and 3416 other…
  5. Qingxiu Dong, Lei Li, Damai Dai, Ce Zheng, Jingyuan Ma, Rui Li, Heming Xia, Jingjing Xu, Zhiyong Wu, Baobao Chang, Xu Sun, Lei Li, and Zhifang Sui. 2024. A survey on in-context learning. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 1107–1128, Miami…
  6. Arushi Goel, Sreyan Ghosh, Jaehyeon Kim, Sonal Kumar, Zhifeng Kong, Sang gil Lee, Chao-Han Huck Yang, Ramani Duraiswami, Dinesh Manocha, Rafael Valle, and Bryan Catanzaro. 2025. Audio flamingo 3: Advancing audio intelligence with fully open large audio language models. In NeurIPS 2025 : Annual Confe… find
  7. Jiliang Hu, Zuchao Li, Baoyuan Qi, Liu Guoming, and Ping Wang. 2026. End-to-end contrastive language-speech pretraining model for long-form spoken question answering. In AAAI 2026. find
  8. Aditya Jain, Fagner Cunha, Michael James Bunsen, Juan Sebastián Cañas, Léonard Pasi, Nathan Pinoy, Flemming Helsing, JoAnne Russo, Marc Botham, Michael Sabourin, Jonathan Fréchette, Alexandre Anctil, Yacksecari Lopez, Eduardo Navarro, Filonila Perez Pimentel, Ana Cecilia Zamora, José Alejandro Ramir… — [arXiv:2406.12452](https://arxiv.org/abs/2406.12452) find
  9. Alexander Johnson, Peter Plantinga, Pheobe Sun, Swaroop Gadiyaram, Abenezer Girma, and Ahmad Emami. 2024. Efficient SQA from long audio contexts: A policy-driven approach. In 25th Annual Conference of the International Speech Communication Association, Interspeech 2024, Kos, Greece, September 1-5, 2… find
  10. KimiTeam, Ding Ding, Zeqian Ju, Yichong Leng, Songxiang Liu, Tong Liu, Zeyu Shang, Kai Shen, Wei Song, Xu Tan, Heyi Tang, Zhengtao Wang, Chu Wei, Yifei Xin, Xinran Xu, Jianwei Yu, Yutao Zhang, Xinyu Zhou, Y. Charles, and 21 others. 2025. Kimi-audio technical report. Preprint, arXiv:2504.18425.
  11. Mikko Lauri, David Hsu, and Joni Pajarinen. 2023a. Partially observable markov decision processes in robotics: A survey. IEEE Transactions on Robotics, 39(1):21–40. find
  12. Mikko Lauri, David Hsu, and Joni Pajarinen. 2023b. Partially observable markov decision processes in robotics: A survey. IEEE Transactions on Robotics, 39(1):21–40. find
  13. Keon Lee, Kyumin Park, and Daeyoung Kim. 2022. Dailytalk: Spoken dialogue dataset for conversational text-to-speech. Preprint, arXiv:2207.01063. — [arXiv:2207.01063](https://arxiv.org/abs/2207.01063) find
  14. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. In Advances in Neural Information Proc…
  15. Chia-Hsuan Li, Szu-Lin Wu, Chi-Liang Liu, and Hung yi Lee. 2018. Spoken squad: A study of mitigating the impact of speech recognition errors on listening comprehension. Preprint, arXiv:1804.00320. — [arXiv:1804.00320](https://arxiv.org/abs/1804.00320) find
  16. Qiwei Li, Teng Xiao, Zuchao Li, Ping Wang, Mengjia Shen, and Hai Zhao. 2025. Dialogue-rag: Enhancing retrieval for llms via node-linking utterance rewriting. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2025, Vienna, Austria,… find
  17. Zheng Lian, Haoyu Chen, Lan Chen, Haiyang Sun, Licai Sun, Yong Ren, Zebang Cheng, Bin Liu, Rui Liu, Xiaojiang Peng, Jiangyan Yi, and Jianhua Tao. 2025a. Affectgpt: A new dataset, model, and benchmark for emotion understanding with multimodal large language models. In Forty-second International Confe… find
  18. Zheng Lian, Haiyang Sun, Licai Sun, Haoyu Chen, Lan Chen, Hao Gu, Zhuofan Wen, Shun Chen, Siyuan Zhang, Hailiang Yao, Bin Liu, Rui Liu, Shan Liang, Ya Li, Jiangyan Yi, and Jianhua Tao. 2025b. OV-MER: towards open-vocabulary multimodal emotion recognition. In Forty-second International Conference on… find
  19. Chyi-Jiunn Lin, Guan-Ting Lin, Yung-Sung Chuang, Wei-Lun Wu, Shang-Wen Li, Abdelrahman Mohamed, Hung-Yi Lee, and Lin-Shan Lee. 2024. Speechdpr: End-to-end spoken passage retrieval for open-domain spoken question answering. In IEEE International Conference on Acoustics, Speech and Signal Processing,… find
  20. Hao Liu, Zhengren Wang, Xi Chen, Zhiyu Li, Feiyu Xiong, Qinhan Yu, and Wentao Zhang. 2025. HopRAG: Multi-hop reasoning for logic-aware retrieval-augmented generation. In Findings of the Association for Computational Linguistics: ACL 2025, pages 1897–1913, Vienna, Austria. Association for Computation… find
  21. Venkatesh Mishra, Bimsara Pathiraja, Mihir Parmar, Sat Chidananda, Jayanth Srinivasa, Gaowen Liu, Ali Payani, and Chitta Baral. 2025. Investigating the shortcomings of LLMs in step-by-step legal reasoning. In Findings of the Association for Computational Linguistics: NAACL 2025, pages 7795–7826, Alb… find
  22. Mingqiu Wang, Izhak Shafran, Hagen Soltau, Wei Han, Yuan Cao, Dian Yu, and Laurent El Shafey. 2024. Retrieval augmented end-to-end spoken dialog models. In IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2024, Seoul, Republic of Korea, April 14-19, 2024, pages 12056–… find
  23. Yijing Wu, SaiKrishna Rallabandi, Ravisutha Srinivasamurthy, Parag Pravin Dakle, Alolika Gon, and Preethi Raghavan. 2023. Heysquad: A spoken question answering dataset. Preprint, arXiv:2304.13689. — [arXiv:2304.13689](https://arxiv.org/abs/2304.13689) find
  24. LLM-Core-Team Xiaomi. 2025. Mimo-audio: Audio language models are few-shot learners.
  25. Jin Xu, Zhifang Guo, Hangrui Hu, Yunfei Chu, Xiong Wang, Jinzheng He, Yuxuan Wang, Xian Shi, Ting He, Xinfa Zhu, Yuanjun Lv, Yongqi Wang, Dake Guo, He Wang, Linhan Ma, Pei Zhang, Xinyu Zhang, Hongkun Hao, Zishan Guo, and 19 others. 2025a. Qwen3-omni technical report. arXiv preprint arXiv:2509.17765.
  26. Kai-Tuo Xu, Feng-Long Xie, Xu Tang, and Yao Hu. 2025b. Fireredasr: Open-source industrial-grade mandarin speech recognition models from encoder-decoder to llm integration. Preprint, arXiv:2501.14350. — [arXiv:2501.14350](https://arxiv.org/abs/2501.14350) find
  27. OpenAI, :, Sandhini Agarwal, Lama Ahmad, Jason Ai, Sam Altman, Andy Applebaum, Edwin Arbus, Rahul K. Arora, Yu Bai, Bowen Baker, Haiming Bao, Boaz Barak, Ally Bennett, Tyler Bertao, Nivedita Brett, Eugene Brevdo, Greg Brockman, Sebastien Bubeck, and 108 others. 2025. gpt-oss-120b. Preprint, arXiv:25…
  28. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R. Narasimhan, and Yuan Cao. 2023. React: Synergizing reasoning and acting in language models. In The Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023. OpenReview.net.
  29. Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. 2022. Robust speech recognition via large-scale weak supervision. Preprint, arXiv:2212.04356.
  30. Chenyu You, Nuo Chen, Fenglin Liu, Shen Ge, Xian Wu, and Yuexian Zou. 2022. End-to-end spoken conversational question answering: Task, dataset and model. In Findings of the Association for Computational Linguistics: NAACL 2022, pages 1219–1232, Seattle, United States. Association for Computational L… find
  31. Timo Schick, Jane Dwivedi-Yu, Roberto Dessi, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. Toolformer: Language models can teach themselves to use tools. In Advances in Neural Information Processing Systems, volume 36, pages 68539–68551. Cu…
  32. Ruihao Shui, Yixin Cao, Xiang Wang, and Tat-Seng Chua. 2023. A comprehensive evaluation of large language models on legal judgment prediction. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 7337–7348, Singapore. Association for Computational Linguistics. find
  33. Quanwei Tang, Sophia Yat Mei Lee, Junshuang Wu, Dong Zhang, Shoushan Li, Erik Cambria, and Guodong Zhou. 2025. A comprehensive graph framework for question answering with mode-seeking preference alignment. In Findings of the Association for Computational Linguistics: ACL 2025, pages 21504– 21523, Vi… find
  34. Fan Yu, Shiliang Zhang, Yihui Fu, Lei Xie, Siqi Zheng, Zhihao Du, Weilong Huang, Pengcheng Guo, Zhijie Yan, Bin Ma, Xin Xu, and Hui Bu. 2022a. M2MeT: The ICASSP 2022 multi-channel multi-party meeting transcription challenge. In Proc. ICASSP. IEEE. find
  35. Fan Yu, Shiliang Zhang, Pengcheng Guo, Yihui Fu, Zhihao Du, Siqi Zheng, Weilong Huang, Lei Xie, Zheng-Hua Tan, DeLiang Wang, Yanmin Qian, Kong Aik Lee, Zhijie Yan, Bin Ma, Xin Xu, and Hui Bu. 2022b. Summary on the ICASSP 2022 multi-channel multi-party meeting transcription grand challenge. In Proc.… find
  36. Wenlin Zhang, Xiangyang Li, Kuicai Dong, Yichao Wang, Pengyue Jia, Xiaopeng Li, Yingyi Zhang, Derong Xu, Zhaocheng Du, Huifeng Guo, Ruiming Tang, and Xiangyu Zhao. 2025. Process vs. outcome reward: Which is better for agentic RAG reinforcement learning. find
  37. Zihan Zhao, Yiyang Jiang, Heyang Liu, Yu Wang, and Yanfeng Wang. 2025. Librisqa: A novel dataset and framework for spoken question answering with large language models. IEEE Trans. Artif. Intell., 6(11):2884–2895. find
  38. Xie Zhifei, Mingbao Lin, Zihang Liu, Pengcheng Wu, Shuicheng Yan, and Chunyan Miao. 2025. Audio-reasoner: Improving reasoning capability in large audio language models. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 23840–23862, Suzhou, China. Associ…

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/tang-2026-don-t-just-listen-try/