基于先验信噪比参数自适应的频域联合语音增强方法
doi: 10.3724/SP.J.1146.2005.00865
A Spectral Domain Compounded Speech Enhancement Algorithm Based on Parameter Adaptive Spectral Method
-
摘要: 该文针对单通道频域语音增强方法存在的问题,提出了一种频域联合语音增强新方法,即将改进的基于先验信噪比频域语音增强方法与迭代谱减法相结合进行语音降噪的方法。实验表明,该方法消噪量大,对语音的损伤小,同时有效地降低了音乐噪声。
-
关键词:
- 语音增强;先验信噪比;谱减法
Abstract: This paper addresses the problem of single-channel speech enhancement using spectral subtraction method. The proposed approach is directed toward combining modified a priori SNR based spectral subtraction algorithm with iterative spectral subtraction algorithm. Experimental results show that the method is quite effective in speech enhancement as it can cancel a large mount of background noise with little distortion to speech signals,and that the method greatly lowered music noise. -
[1] Boll S F. Suppression of acoustic noise in speech using spectral subtraction[J].IEEE Trans. on Acoust, Speech, Signal Processing.1979, 27(2):113-120 [2] Scalart P and Vieira-Filho J. Speech enhancement based on a priori signal to noise estimation. in Proc. ICASSP, Atlanta, 1996: 629-632. [3] Sim B L, Tong Y C, and Chang J S. A parametric formulation of the generalized spectral subtraction method[J].IEEE Trans. on Speech Audio Processing.1998, 6(7):328-337 [4] Ephraim Y and Malah D. Speech enhancement using a minimum mean square error short-time spectral amplitude estimator. IEEE Trans. on Acoust, Speech, Signal Processing, 1984, ASSP-32(12): 1109-1121. [5] Ogata S and Shimamura T. Reinforced spectral subtraction method to enhance speech signal. Electrical and Electronic Technology, 2001, 8(1): 242-245. [6] 杨行峻, 迟惠生. 语音信号数字处理. 北京: 电子工业出版社,1995, 8: 396-399. [7] Marzinzik M and Kollmeier B. Speech pause detection for noise spectrum estimation by tracking power envelope dynamics[J].IEEE Trans. on Speech Audio Processing.2002, 10(2):109-117 [8] Virag N. Single channel speech enhancement based on masking properties of the human auditory system[J].IEEE Trans. on Speech Audio Processing.1999, 7(3):126-137
计量
- 文章访问数: 3278
- HTML全文浏览量: 120
- PDF下载量: 891
- 被引次数: 0