基于理想组合掩蔽的监督性语音增强算法
Supervised Speech Enhancement Algorithm Based on Phase Spectrum Estimation
摘要: 为了解决传统的语音增强算法只对语音幅值谱进行估计，而让语音相位谱保持不变的问题，提出了基于相位谱估计的监督性语音分离算法。首先，对传统的相位补偿理论进行分析，提出了一种同时考虑语音幅值谱和相位谱的理想组合掩码(ICM)，并将其应用到监督性语音增强算法中。经过仿真实验，证实该算法能够有效地抑制背景噪声，并且能够显著地提高语音的可懂性和自动识别率。
Abstract: In order to solve the problem that the traditional speech enhancement algorithms only estimate the speech amplitude spectrum, but make phase spectrum remain unchanged, the supervision of speech separation algorithm based on phase spectrum estimation is proposed. Firstly, after an analysis of the traditional phase compensation, an ideal combination of mask (ICM) considering amplitude spectrum and phase spectrum is proposed and applied to supervised speech enhancement algorithm. The simulation experiment proves the algorithm proposed can not only suppress background noise effectively, but also improve the intelligibility and automatic recognition rate of the speech significantly.
文章引用：李保明, 付小宁. 基于理想组合掩蔽的监督性语音增强算法[J]. 计算机科学与应用, 2018, 8(4): 546-552. https://doi.org/10.12677/CSA.2018.84061

参考文献

[1] 刘文举, 聂帅, 梁山, 等. 基于深度学习语音分离技术的研究现状与进展[J]. 自动化, 2016, 42(6): 819-833.
[2] 高银秋, 邓宗元, 杨震. 数字音频产品中基于人耳听觉感知特性的水印嵌入系统设计[J]. 南京邮电大学学报, 2006, 26(5): 56-64.
[3] Paliwal, K., Wqjcicki, K. and Shannon, B. (2011) The Importance of Phase in Speech Enhancement. Speech Communica-tion, 53, 465-494. [Google Scholar] [CrossRef] 
[4] 薛慧君, 李盛, 路国华, 等. 基于短时相位谱补偿的非接触语音检测增强算法研究[J]. 中国医疗设备报, 2013, 11(28): 12-14.
[5] 孟祥彩, 王中训, 刘伟, 等. 基于相位影响的单声道语音增强改进算法[J]. 通信电声, 2016, 11(40): 59-61.
[6] 王栋, 贾海蓉. 改进相位谱补偿的语音增强算法[J]. 西安电子科技大学学报, 2017, 3(44): 83-88.
[7] Rix, A.W., Beerends, J.G., Hollier, M.P., et al. (2001) Perceptual Evaluation of Speech Quali-ty(PESQ)—A New Method for Speech Quality Assessment of Telephone Networks and Codes. IEEE International Conference on Acoustics, Speech, and Signal Processing, 2001. Proceedings, 749-752.
[8] 王燕南. 基于深度学习的说话人无关单通道语音分离[D]. 合肥: 中国科学技术大学, 2017.
[9] Hinton, G.E., Srivastava, N., Krizhevsky, A., et al. (2012) Improving Neural Net-works by Preventing Co-Adaptation of Feature Detectors. Computer Science, 3, 212-213.
[10] Wang, Y., Han, K. and Wang, D.L. (2013) Exploring Monaural Features for Classification-Based Speech Segregation. IEEE Transaction on Audio Speech & Language Processing, 21, 270-279. [Google Scholar] [CrossRef] 
[11] Taal, C.H., Hendriks, R.C., Heusdens, R., et al. (2011) An Algorithm for In-telligibility Prediction of Time-Frequency Weighted Noisy Speech. IEEE Transactions on Audio Speech & Language Processing, 19, 2125-2136. [Google Scholar] [CrossRef]