Research map / Vision-language and multimodal / Cross-modal learning
VGGSounder: Audio-Visual Evaluations for Foundation Models
Daniil Zverev, Thaddäus Wiedemer, Ameya Prabhu, Matthias Bethge, Wieland Brendel, A. Sophia Koepke
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add human-curated labels to VGGSound ground truth (ablation) | Gemini 1.5 Flash · VGGSounder audible labels (A) | Delta Hit: +29.28 vs +0.48 for auto labels | Improved |
| add human-curated labels to VGGSound ground truth (ablation) | Gemini 1.5 Pro · VGGSounder audible labels (A) | Delta Hit: +28.61 vs +0.31 for auto labels | Improved |
| add human-curated labels to VGGSound ground truth (ablation) | Gemini 2.0 Flash · VGGSounder audible labels (A) | Delta Hit: +8.80 vs +0.22 for auto labels | Improved |
| add visual input to audio-only input | Gemini 1.5 Pro · VGGSounder (no background music) | Hit: 33.43 -> 78.36 | Improved |
| add visual input to audio-only input | Unified-IO 2 · VGGSounder (no background music) | Hit: 54.43 -> 63.36 | Improved |
| add audio input to visual-only input | Unified-IO 2 · VGGSounder (no background music) | Hit: 32.18 -> 63.36 | Improved |
| add audio input to audio-only input | CAV-MAE · VGGSounder (no background music) | Hit: 67.07 -> 67.51 | No clear effect |
| samples with background music | CAV-MAE · VGGSounder audio input | Delta F1: -6.80 | Hurt |
| samples with voice over | CAV-MAE · VGGSounder audio input | Delta F1: -11.93 | Hurt |
| samples with voice over | Gemini 1.5 Flash · VGGSounder audio input | Delta F1: +23.73 | Improved |
| samples with static images | CAV-MAE · VGGSounder audio input | Delta F1: +4.93 | Improved |
| samples with static images | AV-Siam · VGGSounder video input | Delta F1: -9.69 | Hurt |
Closest papers by meaning
- $\texttt{AVROBUSTBENCH}$: Benchmarking the Robustness of Audio-Visual Recognition Models at Test-Time (NeurIPS 2025 · similarity 0.90)
- From Vision to Audio and Beyond: A Unified Model for Audio-Visual Representation and Generation (ICML 2024 · similarity 0.87)
- AVI-Bench: Toward Human-like Audio-Visual Intelligence of Omni-MLLMs (ICML 2026 · similarity 0.87)
- HAVE-Bench: Hierarchical Audio-Visual Evaluation from Perception to Interaction (CVPR 2026 · similarity 0.87)
- V2A-Mapper: A Lightweight Solution for Vision-to-Audio Generation by Connecting Foundation Models (AAAI 2024 · similarity 0.87)
- From Senses to Decisions: The Information Flow of Auditory and Visual Perception in Multimodal LLMs (NeurIPS 2026 · similarity 0.87)
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.
