摘要
目的:研究在强噪声背景条件下增强语音质量的方法,为在复杂条件下获取语音信号奠定基础.方法:在应用小波包分析技术对语音信号进行分解与重构的基础上,对分解后的小波包系数进行尺度,时间2个方面的阈值自适应调节,再对此系数进行重构以实现语音信号的噪声自适应消除.结果:在信噪比为0 dB的强噪声条件下,在0～3000 Hz较宽的频率段上,增强后的语音频谱明显清晰,且各频谱成份更加丰富.结论:本方法能够在强背景噪声条件下对语音信号中的噪声成分进行有效去除.
Abstract
AIM:To explore a novel speech enhancement method based on wavelet packet energy for speech signal acquisition in strong noise background. METHODS: Speech signals were decomposed and reconstructed by application of wavelet packet analysis technology. Both the time and scale thresholds of wavelet packet coefficient were adaptively adjusted so that the wavelet packet coefficient could be reconstructed to accomplish noise adaptive canceling. RESULTS: In condition of strong noise at O dB signal to noise ratio,the spectrum of the enhanced speech was obviously clear and its components were more plentiful than the original speech in a wide range of frequency band of 0 - 3000 Hz. CONCLUSION: Our proposed speech enhancement method is quite effective in removing the noise component of the speech signal in strong noise background.
NETL
NSTL
维普
万方数据