摘要
作为单通道噪声环境下的语音增强,基于短时谱估计的方法如谱减法、维纳滤波法被认为是简单、有效的抑制加性噪声的良好方法。然而,这些方法在抑制噪声的同时会带来高次谐波成分的损失,而高频部分恰恰又是理解语义的关键部分,这样就降低了语音信息的可理解度。本文针对这一情况提出了一种新的方法,即在维纳滤波的基础上,先进行一级时域的梳状滤波,这样可以使语音的浊音部分得到增强并且抑制了维纳滤波后的残留噪声;然后再进行频域梳状滤波,来恢复损失的高频谐波。该方法能更为有效的抑制噪声,同时增强语音的高次谐波,增加语音的可理解度。
As single channel speech enhancement, common methods based on short-time spectral amplitude estimation such as spectral-subtraction and Wiener filtering are considered to be simple and effective for suppressing the additive noise in noisy environments. However, these techniques may distort the high frequency harmonics of speech, which are important to the language understanding. To solve this problem, we propose a new approach： based on Wiener filtering, a comb filter in the time domain is first used. Then another comb filter in frequency domain is implemented, which retrieves the high frequency harmonics that have been compressed. This new method shows an improvement for harmonic enhancement and noise reduction.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2010年第1期26-31,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60802072)
国防科工委基础预研项目(A1320070067)资助
关键词
语音增强
梳状滤波
谐波增强
噪声抑制
speech enhancement
comb filtering
harmonic enhancement
noise suppression
