Choreography Algorithm Based on Hybrid Density Network

Abstract

AbstractChoreography is usually done by professional choreographers, which is highly professional and time-consuming. The development of science and technology is changing the way of artistic creation. The development of motion capture technology and artificial intelligence makes it possible for computer to realize automatic choreography based on music. Two key problems need to be solved in computer music choreography: first, how to get real and novel dance movements without relying on motion capture and handmade; Second, how to use appropriate music and movement features and matching algorithm to enhance the synchronization of music and dance. In order to solve the above two problems, based on the hybrid density network (MDN), the dance matching with the target music is generated through three steps: action generation, action screening and feature matching.KeywordsComputerMusic choreographyMixed density network