虚拟场景中环境声源仿真技术综述

环境声音作为日常生活中分布最为广泛的一类声音,是人们获取外部信息的重要来源.近十几年来,随着用户对虚拟场景真实度要求不断提升,为虚拟场景打造同步、真实的环境音效已成为构建高度沉浸式虚拟环境不可或缺的一部分.其中环境声源仿真作为打造真实感虚拟环境音效的基石,得到了研究人员的广泛关注与探索.与传统的人工声源仿真相比,通过算...

A Survey on Environmental Sound Synthesis for Virtual Scenes

As the most widely distributed type of sound in daily life,the environmental sound is an important source for people to access external information.Over the last decade,with users’increasing requirements for the realism of virtual scenes,it has become an indispensable part of building a highly immersive virtual environment for virtual scenes to create synchronized and realistic environmental sound effects.At present,synthesizing realistic environment sound effects for virtual scenes needs to go through three simulation stages:environment sound synthesis,sound propagation and sound rendering.Among them,environmental sound synthesis,as the cornerstone of creating realistic virtual environment sound effects,has received extensive attention and exploration by researchers.Compared with the traditional artificial sound synthesis,the procedural sound synthesis can not only generate the interactive sound effect,strengthen the immersion of the users in virtual scenes such as games,animation and virtual reality,but also avoid the complex manual work of tuning and editing,which greatly reduces the labor consumption.Most of the existing environmental sound source synthesis work is to explore the specific environmental sound source synthesis method for specific types of sound sources and specific virtual scenes,but the overall environmental sound source synthesis field has not been reviewed and analyzed.In view of this,this paper makes a comprehensive review of the existing environmental sound source synthesis methods used in virtual scenes.This paper classifies and summarizes the environmental sound synthesis methods used in virtual scenes,and makes the overview from multiple perspectives such as research background,application fields,basic principles,and technica l means.Firstly,we start from the discussions on the theoretical research value and the application prospect of environmental sound synthesis and summarize the research field of multi-disciplinary integration.Then,due to that the sound source signal perceived by the human ear is not only related to the physical acoustic principle of the sound source,but also affected by the auditory characteristics of the human ear,we analyze the influence of physical acoustic models and psychoacoustic models on the design and research content of environmental sound synthesis methods.Secondly,according to different construction strategies of sound source models,the existing environmental sound synthesis methods are analyzed from four categories:physical model based,signal model based,hybrid model based and deep learning model based.Among them,the hybrid model based environmental sound synthesis methods combine the sound synthesis methods based on physical model and signal model,while the deep learning model based environmental sound synthesis methods simulate the sound source by constructing the mapping relationship between visual and sound,which is essentially a type of visual-audio cross-modal generation technology.Then,we discuss the development trend,technological innovation and limitations of each different model-based environmental sound synthesis method respectively.Finally,the advantages and disadvantages of different types of method are compared horizontally in terms of algorithm input parameters,algorithm efficiency and simulation result quality,and the main problems and challenges faced by the current environmental sound synthesis field are summarized,and future research directions are prospected.