论文部分内容阅读
本文介绍了一种在频域上对噪声环境中的语言进行处理的方法——减噪声频谱法。利用FFT噪声中的语言信号在频域上展成功率谱,在减去噪声功率谱后,再恢复成语言信号。实践证明,这种方法对消除语言中的宽带噪声有一定的效果,实验中对信噪比为-5dB到+5dB的噪声语言进行了处理,可以抑制近10dB的宽带噪声。
This article describes a method of processing speech in a noisy environment in the frequency domain - noise reduction spectroscopy. The speech signal in FFT noise is used to generate a power spectrum in the frequency domain, and after subtracting the noise power spectrum, the speech signal is restored to a speech signal. Practice has proved that this method to eliminate the broadband noise in the language have some effect, the experiment noise ratio of -5dB to +5 dB noise for processing, can suppress nearly 10dB of broadband noise.