JD.com open-sources JoyAI-Echo long audio-video generation framework
Original title:京东开源JoyAI-Echo框架,进入全球第一梯队!
AI summary
On June 3, JD.com open-sourced the JoyAI-Echo framework to address three major challenges in long video generation: character identity collapse, timbre shifts, and slow generation. The framework maintains character appearance and timbre consistency within 5 minutes through a cross-modal audio-video memory bank; memory-driven post-training combined with DMD technology brings about 7.5x inference acceleration. A new Director Agent supports natural-language conversational local revisions without rerunning the entire video. A lightweight real-time super-resolution module supports two resolutions: 736×1280→1152×1920 and 1472×2560. On the evaluation set, speech content accuracy is 0.8646, leading in multiple user preference metrics. Code and weights have been open-sourced on GitHub.
Key Takeaways
- On June 3,JD.
- com open-sourced the JoyAI-Echo framework to address three major challenges in long video…
- The framework maintains character appearance and timbre consistency within 5 minutes through a…
- memory-driven post-training combined with DMD technology brings about 7.
- 5x inference acceleration.
Quick Facts
- Published
- Updated
- Source
- 公众号:京东JoyAI
- Category
- Model Releases
- AI score
- 72
长视频生成一直被角色崩塌和龟速生成卡死,JoyAI-Echo 开源给出了角色一致性方案和 7.5 倍加速,Director Agent 对话式编辑的思路很先进,做 AI 视频的朋友可以直接去 GitHub 开跑。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: JD.com open-sources JoyAI-Echo long audio-video generation framework. June 5, 2026. https://new.nat6.net/en/news/cmq642x6l0501sl5i3sl7adjt/