88Z AI, 4 June — JD.com launches JoyAI-Echo, a long audio-video generation framework. Through four technical innovations including cross-modal memory banks, it addresses challenges in character consistency, voice stability, and generation speed, with benchmark results showing comprehensive leadership.
At a glance
- JoyAI-Echo addresses character consistency, voice stability, and generation speed challenges through four technical innovations
- Benchmarks show high consistency in character and voice for 5-minute videos, 0.8646 speech accuracy, and 7.5x faster inference
The update
JD.com has officially released the open-source long audio-video generation framework, JoyAI-Echo. The framework introduces four core technical innovations: a cross-modal audio-video memory bank, memory-driven post-training, Director Agent conversational editing, and lightweight real-time super-resolution. These technologies aim to overcome three major industry challenges: maintaining character consistency, controlling voice stability, and improving generation efficiency. Notably, memory-driven post-training increases inference speed by 7.5 times. Benchmark tests show that in five-minute videos generated using this framework, character appearance and voice timbre remain highly consistent, with a speech accuracy rate of 0.8646. All key performance indicators are leading the industry.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- 京东黑板报Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.