基于树莓派4B及Jetson Nano的多路音频采集传输系统  

Multi-channel Audio Acquisition and Transmission System Based on Raspberry Pi 4B and Jetson Nano

作　　者：徐绍峰 陈静轩 刘都鑫 蔡希昌[1] 鲍志平 陈钰婷 刘通[1] 于东池 乔子凌 XU Shao-feng;CHEN Jing-xuan;LIU Du-xin;CAI Xi-chang;BAO Zhi-ping;CHEN Yu-ting;LIU Tong;YU Dong-chi;QIAO Zi-ling(College of Information,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《工业技术创新》2020年第6期40-44,共5页Industrial Technology Innovation

摘　　要：为解决多路音频信号的采集传输及后续信息处理问题,研究一种高性能前端与嵌入式AI后端相结合的系统设计。硬件设计方面,前端采用树莓派4B及6麦克的环形麦克风阵列实现采集及预处理,并通过Wi-Fi进行语音传输;后端处理采用Jetson Nano模块,作为实现语音处理的嵌入式AI平台。软件设计方面,采用C/S架构,进行Linux下的C++编程,其中树莓派作为服务器,实现麦克风信号的采集、编码和网络传输;Jetson Nano模块作为客户机,实现网络通信及实时语音处理功能。测试结果表明,多声道音频处理系统运行稳定,架构合理,具有良好的工程应用潜力。In order to solve issues regarding multi-channel audio signal acquisition,transmission and subsequent processing,the approach of utilizing high-performance front-end architecture with embedded AI back-end is studied.In terms of hardware design:the front-end utilizes Raspberry Pi 4B and 6-microphone ring array for data acquisition and pre-processing,which are transmitted through Wi-Fi;the back-end processing uses Jetson Nano module as an embedded AI platform for voice processing.In terms of software design:the C/S architecture is adopted,and C++programming under Linux is performed,among which,the Raspberry Pi is treated as the server that handles microphone signal acquisition,encoding,and network transmission;the Jetson Nano module functions as the client interface,implementing network communication and real-time voice processing functions.The final test results show that the system runs stably,the structure is reasonable and has good engineering application potential.

关 键 词：嵌入式AI平台 音频传输 树莓派 Jetson Nano模块 多声道音频处理系统 

分 类 号：TN91[电子电信—通信与信息系统]