The Attention Triangle in Audio-Video Models

• URL: http://arxiv.org/abs/2609.03586v1
• Date: Thu, 03 Sep 2026 09:33:27 GMT
• Title: The Attention Triangle in Audio-Video Models
• Abstract summary: Audio-video diffusion models rely on cross-modal attention to coordinate text, sound, and visual content, yet this same mechanism can introduce subtle and systematic semantic leakage.<n>We study these models by probing and analyzing the attention triangle'' comprising the three cross-attention edges connecting the text, audio, and video streams.<n>Our analysis reveals that routing along the audio-video edge is bidirectional: audio can influence video generation, while video can influence audio generation.
• Score: 67.46972888113658
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Audio-video diffusion models rely on cross-modal attention to coordinate text, sound, and visual content, yet this same mechanism can introduce subtle and systematic semantic leakage. We study these models by probing and analyzing the ``attention triangle,'' comprising the three cross-attention edges connecting the text, audio, and video streams, and examine how semantic information is routed across modalities during generation. Our analysis reveals that routing along the audio-video edge is bidirectional: audio can influence video generation, while video can influence audio generation. This edge is shaped by biases encoded in the model's parameters and emerges as a major contributor to leakage: when prompts are in tension with learned priors, cross-modal interactions may override the intended conditioning and reroute semantics toward visually canonical but incorrect outcomes. These effects suggest that semantic artifacts arise not merely from attention spreading beyond its intended target, but from structured, bias-driven interactions along specific pathways. Building on this perspective, we extract attention-derived signals that expose how semantics are distributed and grounded across modalities, and use them as a diagnostic tool to both analyze and deliberately incur leakage under controlled conditions. This enables us to probe the internal dynamics of cross-modal routing and isolate the role of individual interactions. We further leverage these signals to guide inference-time interventions that encourage more consistent cross-modal alignment. Extensive experiments support our analysis and demonstrate improved semantic grounding while preserving generation quality.

Related papers

• Baton: Explicit Semantic Blueprints for Joint Video-Audio Generation [113.24577778437295]
  Baton is a framework that introduces explicit semantic planning into joint video-audio generation.<n>Our key insight is that complementing coarse text guidance with semantically rich, modality-aware planned tokens, can simultaneously restore fine-grained semantic detail.<n> experiments on benchmarks show the effectiveness of Baton both qualitatively and quantitatively.
  arXiv  Detail & Related papers  (2026-05-24T17:55:11Z)
• Separate First, Fuse Later: Mitigating Cross-Modal Interference in Audio-Visual LLMs Reasoning with Modality-Specific Chain-of-Thought [49.53567098922619]
  We propose Separate First, Fuse Later (SFFL) as an audio-visual reasoning framework to reduce cross-modal interference.<n>SFFL enforces modality-specific chain-of-thought reasoning, producing separate audio and visual reasoning traces and integrating evidence for answering.<n> Experiments demonstrate consistent improvements in both accuracy and robustness, yielding an average relative gain of 5.16% on general AVQA benchmarks and 11.17% on a cross-modal hallucination benchmark.
  arXiv  Detail & Related papers  (2026-05-11T02:50:27Z)
• In-Context Audio Control of Video Diffusion Transformers [28.911323185865186]
  This paper introduces In-Context Audio Control of video diffusion transformers (ICAC)<n>We investigate the integration of audio signals for speech-driven video generation within a unified full-attention architecture, akin to FullDiT.<n>We propose a Masked 3D Attention mechanism that constrains the attention pattern to enforce temporal alignment, enabling stable training and superior performance.
  arXiv  Detail & Related papers  (2025-12-21T15:22:28Z)
• Implicit Counterfactual Learning for Audio-Visual Segmentation [50.69377287012591]
  We propose the implicit counterfactual framework (ICF) to achieve unbiased cross-modal understanding.<n>Due to the lack of semantics, heterogeneous representations may lead to erroneous matches.<n>We introduce the multi-granularity implicit text (MIT) involving video-, segment- and frame-level as the bridge to establish the modality-shared space.
  arXiv  Detail & Related papers  (2025-07-28T11:46:35Z)
• Unveiling and Mitigating Bias in Audio Visual Segmentation [9.427676046134374]
  Community researchers have developed a range of advanced audio-visual segmentation models to improve the quality of sounding objects' masks. While masks created by these models may initially appear plausible, they occasionally exhibit anomalies with incorrect grounding logic. We attribute this to real-world inherent preferences and distributions as a simpler signal for learning than the complex audio-visual grounding.
  arXiv  Detail & Related papers  (2024-07-23T16:55:04Z)
• CM-PIE: Cross-modal perception for interactive-enhanced audio-visual video parsing [23.85763377992709]
  We propose a novel interactive-enhanced cross-modal perception method(CM-PIE), which can learn fine-grained features by applying a segment-based attention module. We show that our model offers improved parsing performance on the Look, Listen, and Parse dataset.
  arXiv  Detail & Related papers  (2023-10-11T14:15:25Z)
• Cross-Modal Attention Consistency for Video-Audio Unsupervised Learning [141.38505371646482]
  Cross-modal correlation provides an inherent supervision for video unsupervised representation learning. This paper introduces a pretext task, Cross-Modal Attention Consistency (CMAC), for exploring the bidirectional local correspondence property. CMAC aims to align the regional attention generated purely from the visual signal with the target attention generated under the guidance of acoustic signal.
  arXiv  Detail & Related papers  (2021-06-13T07:41:15Z)
• Unsupervised Sound Localization via Iterative Contrastive Learning [106.56167882750792]
  We propose an iterative contrastive learning framework that requires no data annotations. We then use the pseudo-labels to learn the correlation between the visual and audio signals sampled from the same video. Our iterative strategy gradually encourages the localization of the sounding objects and reduces the correlation between the non-sounding regions and the reference audio.
  arXiv  Detail & Related papers  (2021-04-01T07:48:29Z)
• Look, Listen, and Attend: Co-Attention Network for Self-Supervised Audio-Visual Representation Learning [17.6311804187027]
  An underlying correlation between audio and visual events can be utilized as free supervised information to train a neural network. We propose a novel self-supervised framework with co-attention mechanism to learn generic cross-modal representations from unlabelled videos. Experiments show that our model achieves state-of-the-art performance on the pretext task while having fewer parameters compared with existing methods.
  arXiv  Detail & Related papers  (2020-08-13T10:08:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.