---
格式版本: 2
标题: "TurboT2VA: Fast Large-Scale Text-to-Video-Audio Generation via Score-Regularized Consistency Distillation"
原文链接: "https://arxiv.org/abs/2608.24674"
发布日期: "2026-08-25"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Tue, 25 Aug 2026 15:09:02 UTC (6,994 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-26T21:53:10+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-26T21:50:39+08:00"
入库时间: "2026-08-26T13:53:10.337Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=Scale-up&searchtype=all"
匹配关键词:
  - "Scale-up"
  - "deployment"
  - "latency"
相关厂家:
  - "NVIDIA"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 37
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文主线是19B视频音频生成模型的蒸馏与单卡推理加速，不是超节点或机架级AI基础设施。arXiv原始预印本可追溯但尚非正式评审成果；固定知识库未见同项记录，但不能据此认定首次出现。新增事实包括四步蒸馏、W8A8/算子融合/稀疏注意力，以及H20单卡延迟由318.74秒降至5.83秒；没有客户、量产或规模部署。命中“应用与模型效率”硬否决，相关机制和指标主要服务单一生成模型，不能形成可独立复用的机架级基础设施方案。"
AI质检模型: "gpt-5.6-sol"
AI质检时间: "2026-08-26T21:53:21+08:00"
AI主题相关性: 1
AI来源权威性: 11
AI新颖性: 14
AI技术细节: 4
AI商业部署信号: 0
AI完整性: 7
AI评分提示词版本: "v17-精简生产版"
AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d"
AI评分知识库版本: "knowledge_base_v1-20260819+runtime.3"
AI评分知识库SHA256: "dbc02c7552b478ae5aae514533e58a5de9ce72d2911e4aaac0f744c08178e4a6"
AI评分知识库检索词: "[\"Scale-up\",\"NVIDIA\",\"TurboT2VA\",\"PDF\",\"arxiv.org/pdf/2608.24674\",\"HTML\",\"arxiv.org/html/2608.24674v1\",\"T2VA\",\"v1\",\"LTX-2\",\"W8A8\",\"H20\"]"
AI评分知识库命中: "[{\"id\":\"july-correct-0111\",\"title\":\"StrataCL: Fabric-Native Communication Library for Production Supernodes\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"PDF\",\"HTML\",\"v1\"],\"rank\":-10.496705137689483},{\"id\":\"july-correct-0010\",\"title\":\"Schneider Electric and AMD release first Helios platform reference design to accelerate AI Factory deployment\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"PDF\",\"v1\"],\"rank\":-7.2252356915138165},{\"id\":\"july-correct-0062\",\"title\":\"Why AI Racks Need an Open Signal Conditioning Standard\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"Scale-up\",\"NVIDIA\",\"PDF\",\"v1\"],\"rank\":-7.192080541549073},{\"id\":\"historical-jun-010\",\"title\":\"爱建证券-电子行业专题报告：Vera Rubin量产提速，RTX Spark打开终端AI新空间-260608.pdf\",\"sourceType\":\"curated_item\",\"time\":\"2026-06\",\"matchedTerms\":[\"PDF\"],\"rank\":-6.792800214723912},{\"id\":\"july-correct-0081\",\"title\":\"AAI 2026: 6th Gen AMD EPYC Server CPUs Power the Agentic Data Center\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"NVIDIA\",\"PDF\",\"HTML\"],\"rank\":-6.51721352712259}]"
AI摘要: "TurboT2VA 提出一种蒸馏与推理框架，用于加速 19B 参数的联合文本生成视频与音频模型，通过分模态归一化和渐进式课程解决大规模蒸馏中的模态失衡与质量多样性权衡。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-08-27T08:12:49.951Z"
采集批次: "2026年8月26日20点37分07秒"
采集批次ID: "20260826-203707-934"
去重键: "https://arxiv.org/abs/2608.24674"
---

## Computer Science > Computer Vision and Pattern Recognition

## Title:TurboT2VA: Fast Large-Scale Text-to-Video-Audio Generation via Score-Regularized Consistency Distillation

[View PDF](https://arxiv.org/pdf/2608.24674) [HTML (experimental)](https://arxiv.org/html/2608.24674v1)

> Abstract:Joint text-to-video-audio generation produces synchronized visual and acoustic content, but the long sampling trajectories and heterogeneous multimodal computation of large models make inference prohibitively expensive. We present TurboT2VA, a distillation and inference framework for accelerating a 19B-parameter joint video-audio model. Large-scale T2VA distillation is challenged by modality-imbalanced optimization, the difficulty of continuous-time consistency training at scale, and the quality--diversity trade-off. TurboT2VA addresses these issues with per-modality normalization and a progressive curriculum comprising discrete consistency warm-up, continuous consistency refinement, and joint consistency--distribution matching. The curriculum first establishes a stable, diverse generation trajectory and only then introduces distribution-level refinement. On LTX-2, four-step distillation reduces generator latency from 50.52s to 2.51s at the standard evaluation resolution of 512 $\times$ 768, achieving a 20.1 $\times$ speedup while maintaining strong visual quality, audio fidelity, diversity, and video-audio synchronization. We further develop an architecture-aware inference stack that combines guarded W8A8 and fused operators, padded-text compaction, and modality-aware sparse attention while preserving dense cross-modal and text-conditioning paths. Under the high-resolution deployment setting at 1024 $\times$ 1792, the complete stack reduces generator latency from 318.74s to 5.83s on one NVIDIA H20, achieving a 54.67 $\times$ generator-only speedup. Inference code and generation demos are available at [this https URL](https://github.com/thu-ml/TurboDiffusion/tree/main/turbot2va).

| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| --- | --- |
| Cite as: | [arXiv:2608.24674](https://arxiv.org/abs/2608.24674) \[cs.CV\] |
|  | (or [arXiv:2608.24674v1](https://arxiv.org/abs/2608.24674v1) \[cs.CV\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.24674](https://doi.org/10.48550/arXiv.2608.24674) |

## Submission history

From: Yuxiang Liu \[[view email](https://arxiv.org/show-email/0fb96515/2608.24674)\]  
**\[v1\]** Tue, 25 Aug 2026 15:09:02 UTC (6,994 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.24674) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))