VA-Judger: Reward Modeling from Human Preference Feedback for Joint Video-Audio Generation

• URL: http://arxiv.org/abs/2608.18607v2
• Date: Thu, 20 Aug 2026 15:01:06 GMT
• Title: VA-Judger: Reward Modeling from Human Preference Feedback for Joint Video-Audio Generation
• Abstract summary: We propose VA-Judger, a chain-of-thought omni-reward model for joint video-audio generation.<n>We show VA-Judger outperforms metric baselines in predicting human preferences on both in-domain and out-of-domain evaluations.
• Score: 79.34043102155964
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Using reinforcement learning to post-train joint video-audio generation models requires a reward signal. Existing methods construct this reward by combining metrics for individual quality dimensions, including audio quality, visual fidelity, and synchronization. However, these metrics evaluate perceptual dimensions separately and fail to capture the overall semantic and temporal coherence among the text prompt, video, and audio that shapes human preferences. Optimizing models against these metrics encourages reward hacking, generating video-audio content that achieves high scores on these metrics yet appears incoherent or unfaithful to human viewers. To address this problem, we first construct a large-scale human-preference dataset VAPref-10K for joint video-audio generation, comprising 9K prompts and 10.3K fine-grained paired comparisons from open-source generation models. We also introduce the VA-Judger-Bench benchmark with both in-domain and out-of-domain model comparisons to evaluate whether reward models truly align with human preferences. We further propose VA-Judger, a chain-of-thought omni-reward model for joint video-audio generation. In particular, VA-Judger first learns from pairs with clear quality gaps to establish structured output and coarse preference discrimination, then distills reliable preference explanations for harder near-quality comparisons via rejection sampling verified against human annotations, and finally performs dimension-wise reinforcement learning that decomposes human feedback into individual quality dimensions for denser reward signals than a single binary preference label. Experiments show that VA-Judger outperforms metric baselines in predicting human preferences on both in-domain and out-of-domain evaluations. Using its human-aligned rewards for post-training audio-video generation model also yields significant improvements in generation quality.

Related papers

• Aligning Human Sense: Calibrated Distributional Reward Learning for Video Generation [8.918851532959815]
  Video generation is central to AI-powered content creation.<n>Despite progress in visual quality, three key challenges remain.<n>First, the reliability of reward signals is constrained by the quality of human preference data.<n>Second, standard scalar reward models collapse multi-aspect human preferences into a single value.<n>Third, in policy optimization, the widely adopted KL divergence imposes primarily local constraints.
  arXiv  Detail & Related papers  (2026-08-16T07:50:30Z)
• Inference-Time Scaling for Joint Audio-Video Generation [38.09471807128537]
  Joint audio-video generation models often require substantial training resources to improve fidelity.<n>Inference-Time Scaling is a promising training-free alternative in single-modality domains.<n>We present the first comprehensive study of ITS for joint audio-video generation.
  arXiv  Detail & Related papers  (2026-06-02T05:41:41Z)
• JavisDiT++: Unified Modeling and Optimization for Joint Audio-Video Generation [112.614973927778]
  Joint audio-video generation (JAVG) produces synchronized and semantically aligned sound and vision from textual descriptions.<n>This paper presents JavisDiT++, a framework for unified modeling and optimization of JAVG.<n>Our model achieves state-of-the-art performance merely with around 1M public training entries.
  arXiv  Detail & Related papers  (2026-02-22T12:44:28Z)
• High-Quality Sound Separation Across Diverse Categories via Visually-Guided Generative Modeling [65.02357548201188]
  We propose DAVIS, a Diffusion-based Audio-VIsual Separation framework that solves the audio-visual sound source separation task through generative learning.<n>Our framework operates by synthesizing the desired separated sound spectrograms directly from a noise distribution, conditioned concurrently on the mixed audio input and associated visual information.
  arXiv  Detail & Related papers  (2025-09-26T08:46:00Z)
• Multi-Metric Preference Alignment for Generative Speech Restoration [15.696247605348383]
  We propose a multi-metric preference alignment strategy for generative models.<n>We observe consistent and significant performance gains across three diverse generative paradigms.<n>Our aligned models can serve as powerful ''data annotators'', generating high-quality pseudo-labels.
  arXiv  Detail & Related papers  (2025-08-24T07:05:10Z)
• VideoDPO: Omni-Preference Alignment for Video Diffusion Generation [48.36302380755874]
  Direct Preference Optimization (DPO) has demonstrated significant improvements in language and image generation.<n>We propose a VideoDPO pipeline by making several key adjustments.<n>Our experiments demonstrate substantial improvements in both visual quality and semantic alignment.
  arXiv  Detail & Related papers  (2024-12-18T18:59:49Z)
• 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)
• MM-Diffusion: Learning Multi-Modal Diffusion Models for Joint Audio and Video Generation [70.74377373885645]
  We propose the first joint audio-video generation framework that brings engaging watching and listening experiences simultaneously. MM-Diffusion consists of a sequential multi-modal U-Net for a joint denoising process by design. Experiments show superior results in unconditional audio-video generation, and zero-shot conditional tasks.
  arXiv  Detail & Related papers  (2022-12-19T14:11:52Z)
• Audio-visual speech enhancement with a deep Kalman filter generative model [0.0]
  We present an audiovisual deep Kalman filter (AV-DKF) generative model which assumes a first-order Markov chain model for the latent variables. We develop an efficient inference methodology to estimate speech signals at test time.
  arXiv  Detail & Related papers  (2022-11-02T09:50:08Z)
• Switching Variational Auto-Encoders for Noise-Agnostic Audio-visual Speech Enhancement [26.596930749375474]
  We introduce the use of a latent sequential variable with Markovian dependencies to switch between different VAE architectures through time. We derive the corresponding variational expectation-maximization algorithm to estimate the parameters of the model and enhance the speech signal.
  arXiv  Detail & Related papers  (2021-02-08T11:45:02Z)
• NAViDAd: A No-Reference Audio-Visual Quality Metric Based on a Deep Autoencoder [0.0]
  We propose a No-Reference Audio-Visual Quality Metric Based on a Deep Autoencoder (NAViDAd) The model is formed by a 2-layer framework that includes a deep autoencoder layer and a classification layer. The model performed well when tested against the UnB-AV and the LiveNetflix-II databases.
  arXiv  Detail & Related papers  (2020-01-30T15:40:08Z)

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.