Tencent Hybrid releases open source voice digital human model HunyuanVideo-Avatar

Tencent hybrid open source released voice digital human model HunyuanVideo-Avatar, only a picture and a piece of audio can be generated character natural speech or singing video; the model supports head and shoulders, half-body and full-body scene, can understand the image environment and audio emotion, automatically generate natural expression, lip synchronization and full-body action; technology has been applied to Tencent music products, applicable to short-term video creation, e-commerce advertising and other scenes, supporting multi-style and two-person interactive scenes. The technology has been applied to Tencent music products and is suitable for short videos, e-commerce advertisements and other scenarios, supporting multi-style, multi-species and two-player interactive scenarios.