Aligned Better, Listen Better For Audio-Visual Large Language Models

Yuxin Guo, Shuailei Ma, Shijie Ma, Xiaoyi Bao, Chen-Wei Xie, Kecheng Zheng, Tingyu Weng, Siyang Sun, Yun Zheng, Wei Zou

Keywords: Audio-Visual Learning, Multimodal Large Language Models

TL;DR: We introduce an audio-visual multi-scale adapter that can extract and merge spatial information from both modalities at multiple scales, thereby enhancing feature interaction and spatial alignment between modalities.

Abstract: Audio is essential for multimodal video understanding. On the one hand, video inherently contains audio and audio supplies complementary information to the visual modality. Besides, video large language models (Video-LLMs) can encounter many audio-centric settings. However, existing Video-LLMs and Audio-Visual Large Language Models (AV-LLMs) exhibit deficiencies in exploiting audio information, leading to weak understanding and hallucination. To solve the issues, we delve into the model architecture and data aspects. (1) From the architectural perspective, we propose a fine-grained AV-LLM, namely Dolphin. The concurrent alignment of audio and visual modalities in both temporal and spatial dimensions ensures a comprehensive and accurate understanding of videos. Specifically, we devise an audio-visual multi-scale adapter for multi-scale information aggregation, which achieves spatial alignment. For temporal alignment, we propose audio-visual interleaved merging. (2) From the data perspective, we curate an audio-visual caption \& instruction-tuning dataset, called AVU. It comprises 5.2 million diverse, open-ended data tuples (video, audio, question, answer) and introduces a novel data partitioning strategy. Extensive experiments show our model not only achieves remarkable performance in audio-visual understanding, but also mitigates hallucinations. Our codes and dataset will be made publicly available.

Primary Area: foundation or frontier models, including LLMs

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 3083