On October 5, 2025, the Chinese mainland unveiled its first national standard for customer service virtual digital humans, marking a milestone in AI-driven interactions. The Information technology – General technical requirements for customer service virtual digital human standard establishes a unified framework for the research, production, and deployment of digital assistants across finance, government affairs, education, and more.
Key technical specifications
- Avatar generation: 2D avatars must capture complete facial details; 3D models require at least 200,000 polygons for lifelike rendering.
- Multi-modal interaction: Support for voice, gesture, and body movement, with operational maintenance like keyword tracking and corpus updates.
- Lip-sync accuracy: Minimum 90% to ensure speech and mouth movements stay in harmony.
- Gesture and movement success rates: At least 90% accuracy for natural body language communication.
- Emotional intelligence: Recognize joy, sadness, and anxiety with an 80% success rate, responding with expressive visuals and emotional speech synthesis.
- Response time and understanding: Under two seconds for speech interactions and 85% accuracy in semantic understanding.
Looking ahead: flexible and future-ready
The standard applies to existing 2D and 3D digital human products and paves the way for integrating emerging AI-generated content. With supporting testing methods in development, enterprises across sectors will soon have a unified benchmark to identify issues and optimize their virtual assistants quickly.
At a time when digitally savvy global citizens demand seamless, context-aware service, this standard signals the Chinese mainland's commitment to elevating virtual human experiences and setting the stage for more empathetic, intelligent AI interactions worldwide.
Reference(s):
China issues first national standard for virtual digital humans
cgtn.com
