Quality Over Quantity? LLM-Based Curation for a Data-Efficient Audio-Video Foundation Model

• URL: http://arxiv.org/abs/2503.09205v2
• Date: Thu, 13 Mar 2025 18:37:01 GMT
• Title: Quality Over Quantity? LLM-Based Curation for a Data-Efficient Audio-Video Foundation Model
• Authors: Ali Vosoughi, Dimitra Emmanouilidou, Hannes Gamper,
• Abstract summary: AVVA scores and selects high-quality training clips using Whisper for audio and DINOv2 for video within a dual-encoder contrastive learning framework.<n>Trading data quantity for data quality improves performance, yielding respective top-3 accuracy increases of 47.8, 48.4, and 58.0 percentage points on AudioCaps, VALOR, and VGGSound.
• Score: 11.010635593271045
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Integrating audio and visual data for training multimodal foundational models remains challenging. We present Audio-Video Vector Alignment (AVVA), which aligns audiovisual (AV) scene content beyond mere temporal synchronization via a Large Language Model (LLM)-based data curation pipeline. Specifically, AVVA scores and selects high-quality training clips using Whisper (speech-based audio foundation model) for audio and DINOv2 for video within a dual-encoder contrastive learning framework. Evaluations on AudioCaps, VALOR, and VGGSound demonstrate that this approach can achieve significant accuracy gains with substantially less curated data. For instance, AVVA yields a 7.6% improvement in top-1 accuracy for audio-to-video retrieval on VGGSound compared to ImageBind, despite training on only 192 hours of carefully filtered data (vs. 5800+ hours). Moreover, an ablation study highlights that trading data quantity for data quality improves performance, yielding respective top-3 accuracy increases of 47.8, 48.4, and 58.0 percentage points on AudioCaps, VALOR, and VGGSound over uncurated baselines. While these results underscore AVVA's data efficiency, we also discuss the overhead of LLM-driven curation and how it may be scaled or approximated in larger domains. Overall, AVVA provides a viable path toward more robust, text-free audiovisual learning with improved retrieval accuracy.

Related papers

• SAVEn-Vid: Synergistic Audio-Visual Integration for Enhanced Understanding in Long Video Context [19.224601064352846]
  We introduce SAVEn-Vid, the first-ever long audio-visual video dataset comprising over 58k audio-visual instructions. We present AVBench, a benchmark containing 2,500 QAs designed to evaluate models on enhanced audio-visual comprehension tasks within long video. Experiments demonstrate that SAVEnVideo outperforms the best Video-LLM by 3.61% on the zero-shot long video task (Video-MME) and surpasses the leading audio-visual LLM by 1.29% on the zero-shot audio-visual task (Music-AVQA)
  arXiv  Detail & Related papers  (2024-11-25T09:22:13Z)
• Audio-visual training for improved grounding in video-text LLMs [1.9320359360360702]
  We propose a model architecture that handles audio-visual inputs explicitly. We train our model with both audio and visual data from a video instruction-tuning dataset. For better evaluation of audio-visual models, we also release a human-annotated benchmark dataset.
  arXiv  Detail & Related papers  (2024-07-21T03:59:14Z)
• SAVE: Segment Audio-Visual Easy way using Segment Anything Model [0.0]
  This study presents a lightweight approach, SAVE, which efficiently adapts the pre-trained segment anything model (SAM) to the AVS task. Our proposed model achieves effective audio-visual fusion and interaction during the encoding stage.
  arXiv  Detail & Related papers  (2024-07-02T07:22:28Z)
• Frieren: Efficient Video-to-Audio Generation Network with Rectified Flow Matching [51.70360630470263]
  Video-to-audio (V2A) generation aims to synthesize content-matching audio from silent video.<n>We propose Frieren, a V2A model based on rectified flow matching.<n>Experiments indicate that Frieren achieves state-of-the-art performance in both generation quality and temporal alignment.
  arXiv  Detail & Related papers  (2024-06-01T06:40:22Z)
• Text-to-feature diffusion for audio-visual few-shot learning [59.45164042078649]
  Few-shot learning from video data is a challenging and underexplored, yet much cheaper, setup. We introduce a unified audio-visual few-shot video classification benchmark on three datasets. We show that AV-DIFF obtains state-of-the-art performance on our proposed benchmark for audio-visual few-shot learning.
  arXiv  Detail & Related papers  (2023-09-07T17:30:36Z)
• AdVerb: Visually Guided Audio Dereverberation [49.958724234969445]
  We present AdVerb, a novel audio-visual dereverberation framework. It uses visual cues in addition to the reverberant sound to estimate clean audio.
  arXiv  Detail & Related papers  (2023-08-23T18:20:59Z)
• Unraveling Instance Associations: A Closer Look for Audio-Visual Segmentation [18.001730255429347]
  Audio-visual segmentation (AVS) is a challenging task that involves accurately segmenting sounding objects based on audio-visual cues. We propose a new cost-effective strategy to build challenging and relatively unbiased high-quality audio-visual segmentation benchmarks. Experiments conducted on existing AVS datasets and on our new benchmark show that our method achieves state-of-the-art (SOTA) segmentation accuracy.
  arXiv  Detail & Related papers  (2023-04-06T09:54:06Z)
• AVFormer: Injecting Vision into Frozen Speech Models for Zero-Shot AV-ASR [79.21857972093332]
  We present AVFormer, a method for augmenting audio-only models with visual information, at the same time performing lightweight domain adaptation. We show that these can be trained on a small amount of weakly labelled video data with minimum additional training time and parameters. We also introduce a simple curriculum scheme during training which we show is crucial to enable the model to jointly process audio and visual information effectively.
  arXiv  Detail & Related papers  (2023-03-29T07:24:28Z)
• Contrastive Audio-Visual Masked Autoencoder [85.53776628515561]
  Contrastive Audio-Visual Masked Auto-Encoder (CAV-MAE) Our fully self-supervised pretrained CAV-MAE achieves a new SOTA accuracy of 65.9% on VGGSound.
  arXiv  Detail & Related papers  (2022-10-02T07:29:57Z)
• Robust Self-Supervised Audio-Visual Speech Recognition [29.526786921769613]
  We present a self-supervised audio-visual speech recognition framework built upon Audio-Visual HuBERT (AV-HuBERT) On the largest available AVSR benchmark dataset LRS3, our approach outperforms prior state-of-the-art by 50% (28.0% vs. 14.1%) using less than 10% of labeled data. Our approach reduces the WER of an audio-based model by over 75% (25.8% vs. 5.8%) on average.
  arXiv  Detail & Related papers  (2022-01-05T18:50:50Z)
• VGGSound: A Large-scale Audio-Visual Dataset [160.1604237188594]
  We propose a scalable pipeline to create an audio dataset from open-source media. We use this pipeline to curate the VGGSound dataset consisting of more than 210k videos for 310 audio classes. The resulting dataset can be used for training and evaluating audio recognition models.
  arXiv  Detail & Related papers  (2020-04-29T17:46:54Z)

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.