嵌入式数字录音中背景噪声抑制方法研究

本文分析比较了语音和各种背景噪声的特性,采用滤波法、能量法与自相关函数法三者相结合的方法,提出了一种背景噪声检测和噪声抑制的算法,分析了算法的时间复杂度,建立了适用于嵌入式录音系统的噪声去除方案;为了对该方案进行验证,针对机车上的录音样本进行了噪声检测去除的实验,实验结果表明:通过噪声抑制后语音段周期噪声得以去除,提高了录音的音质,另外语音和纯噪声得以准确的区分,并去除纯噪声,能够减少录音的存储量;最后给出了采用数字信号处理器的嵌入式系统具体设计.

Research on Background Noise Suppression Method of Embedded Digital Recording

This paper analyzed and compared the characteristic of speech and background noise. The technique of combining filtering method, energy method and auto correlation function method is adopted, and the algorithm of background noise detection and suppression is given. The time complexity analysis of the algorithm is also given. A sort of noise-eliminating scheme that is applicable in embedded recording system is proposed. In order to test the scheme, an experiment of noise eliminating for locomotive recording samples is performed. The results shown that noise can be eliminated effectively by the means of noise suppression, the quality of the voice is improved, and the amount of the data recording is also reduced greatly. In the end, a practical embedded design that realizes those functions and includes a digital signal processor is introduced.