基于TMS320F2812的声信号采集与特征提取设计

Design of Sound Acquisition and Feature Extraction Based on TMS320F2812

介绍了系统的总体设计并给出前端信号调理电路的详细原理,以声信号为研究对象,以DSP芯片TMS320F2812为核心,使用过采样算法和小波阈值去噪法实现了对声信号的实时采集与滤波处理,并对声信号构建AR模型采用Yule-Walker法完成了特征提取,满足了如目标识别这种对采集处理实时性要求较高的系统的需求。另外,对采集的数据和提取的特征还可以通过SCI口传输到上位机,便于分析和处理。实验表明,该设计切实可行,应具有较好的应用前景。

This paper introduced the collectivity design of system and gave the anterior adjusting circuit in detail. The design took sound signal as research object, took DSP chip TMS320F2812 as the core, used the methods of over-sample arithmetic, wavelet threshold denoising to realize real-time acquisition and filtering processing, completed the feature extraction by constructing AR model for sound signal with the method of Yule-Walker. It satisfies the high need of real-time feature for acquisition processing just as target recognition system. In addition, the system can transmit origin data and feature to PC from SCI which is convenient for analyzing and processing. The experiment shows that the design is feasible and has a better application prospect.