基于SystemVue的声信号分离仿真系统的建模与实现

Modeling and Realization of Acoustic Signal Separation Simulation System Based on System Vue

针对现代声信号分离过程中的实际需求,按照信号分离的具体模式以及声传感器在实际工作中的网络布局,提出一种基于System Vue仿真平台下的声信号分离仿真系统的建模方法,在Fast ICA算法的基础上进行改进,提出一种基于时频单源主导区域检测的信号分离算法,使得在未知声信号源数目的条件下得到源信号波形的准确估计,为声传感器的信号提取与处理提供新思路。利用System Vue仿真软件对声传感器网络进行建模,同时形成目标的模拟信息,对信号的传播环境进行详细分析。在仿真过程中,将基于时频单源主导区域检测的信号分离算法与传统的引号分离算法进行对比以验证其可行性和优越性,仿真系统对信号分离速度以及信噪比两方面进行测试,结果表明该建模方案的正确性以及信号分离算法的优越性。

Aiming at the practical requirements of blind separation process of acoustic signals, a modeling method of acoustic signals separation simulation system based on System Vue simulation platform is proposed according to specific mode of signal separation and acoustic sensor network layout in the practical work. It provides a novel signal separation algorithm based on signal single source detection which is the improvement of FastlCA algorithm. It also can obtain the source acoustic＇ signals without the information of signal source numbers. The SystemVue software is used to model the net of acoustic sensors, and the target simulation information is generated at the same time. In the simulation, the comparison between proposed signal separation algorithm and traditional algorithms is to prove the feasibility and superiority of the proposed algorithm. It is approved that the modeling scheme is correct and realizable by the simulation experiment.