Research on Speech-based Digital Human Driving System Based on Unreal Engine

Abstract

This paper proposed and realized a speech-based digital human driving system in Unreal engine. Based on blendshapes, this paper proposed a micro-expression driving algorithm to realize the change of expressions dynamically. In order to drive lip shapes efficiently, shape libraries are used to cache basic facial data instead of storing animations, which can reduce additional overhead of hardware resources, and can also achieve reuse and expansion of data. Moreover, this paper introduces the proxy middleware into the architecture, which can reduce bandwidth consumption. This system can be applied in intelligent services such as tour guiding, educational counseling, exhibition interpreting and other scenes, which does not affect the experiences of activities, but can greatly reduce the cost in terms of transportation, time and financial resources, and has broad application prospects.