After inputting speech and facial images, the system employs three self-developed core algorithms to achieve speech generation of high-precision 3D expression parameters, generation of natural and stable head motion parameters, and parameter-driven facial video generation. Then outputs talking-face videos with accurate lip-sync, natural head movements, and blinks, thus achieving high-precision audio-visual synchronization and realistic motion.

Technology provider:Xi'an Jiaotong University Technology Transfer Center