Carl Tu
Xiaoice Company announced the official launch of its new “Zero-Shot” digital human technology (Zero-shot Xiaoice Neural Rendering, Zero-XNR) and simultaneously introduced a new, universally accessible digital employee product based on Z-XNR technology, further enriching the Xiaoice AI digital employee product line.
Zero-Shot Technology and Large Model Integration
The new technology adheres to the “Digital Human + Large Model” approach that the Xiaoice AI digital employee product line consistently follows. Compared to other existing technologies, this new advancement leverages a trillion-parameter model base and digital human interaction suites built on large models. It compresses the training data required for digital humans to “seconds,” making customization almost instantaneous. The generated digital humans can be used directly for real-time interaction, a global first. The new technology maintains the ultra-high-definition standards of Xiaoice AI digital employee products, achieving lifelike recreations of human employees’ appearances and voices.
Key Upgrades to Xiaoice Digital Human Technology
- New Z-XNR Technology
The fusion framework of Z-XNR “Zero-Shot” digital human technology and TTS voice large models allows for high-quality image and voice replication within seconds. The unique data training and inference algorithms ensure high-quality replication, personalized visual expression, and immediate availability. - New Trillion-Parameter Model Base and Agent Construction Framework
Based on the trillion-parameter model base, the upgraded hybrid base architecture’s Agent construction framework builds a comprehensive platform for developing and enhancing digital employee capabilities—the Digital Brain Platform. This equips the entire series of digital employees with powerful and diverse professional interaction suites, enabling them to better understand customers and enterprise knowledge and business, with interactions precisely tailored to native business workflows. - New Enhanced Audio-Visual Transmission System
Utilizing a self-developed engine based on WebRTC, it provides ultra-high-definition video streaming capabilities, with telecom-grade high-quality, low-latency audio-visual transmission and robust network packet loss resistance. It ensures precise audio-visual synchronization and can be integrated with…
This text provides an overview of the latest technological advancements in Xiaoice’s AI digital employee solutions, highlighting significant improvements in real-time digital human interactions and enhanced communication capabilities.
