单信道语音增强中先验信噪比参数估计算法的对比分析
Comparison and Analysis of a Priori SNR Parameter Estimation Algorithm in Single Channel Speech Enhancement
摘要: 先验信噪比参数估计的准确性是决定噪声背景下语音增强系统输出性能的关键因素。直接判决(Decision-Directed, DD)技术是先验信噪比估计体系中最为简易直接的算法，后续算法多为对此技术的进一步优化或改进。本文对常用的直接判决算法、两步噪声消除(Two-step Noise Reduction, TSNR)算法、改进的两步噪声消除(Modified TSNR, MTSNR)算法以及融合耦合因子(Convex Combination, CC)算法等四种先验信噪比技术进行了对比分析，给出了各种算法的基本设计原理，并从理论分析和实验仿真两个方面讨论了四种算法的输出性能及其优缺点。
Abstract: The accuracy of a priori signal-to-noise ratio parameter estimation is a key factor in determining the output performance of the speech enhancement system in the noise background. The Deci-sion-Directed (DD) technique is the simplest and straightforward algorithm in the a priori sig-nal-to-noise ratio estimation system. Subsequent algorithms are mostly the further optimization or improvement of this technique. This paper deals with four commonly used direct decision algo-rithms, Two-step Noise Reduction (TSNR) algorithm, Modified Two-step Noise Cancellation (MTSNR) algorithm, and Convex Combination (CC) algorithm. A priori signal to noise ratio technology was compared and analyzed. The basic design principles of various algorithms were given. The output performance and advantages and disadvantages of the four algorithms were discussed from the aspects of theoretical analysis and experimental simulation.
文章引用：陈晨, 高颖, 刘伟, 韩蕊蕊, 张硕. 单信道语音增强中先验信噪比参数估计算法的对比分析[J]. 电路与系统, 2018, 7(2): 25-35. https://doi.org/10.12677/OJCS.2018.72004

参考文献

[1] Schwerin, B. and Paliwal, K. (2014) Using STFT Real and Imaginary Parts of Modulation Signals for MMSE-Based Speech Enhancement. Speech Communication, 58, 49-68.
[Google Scholar] [CrossRef] 
[2] Fang, Y., Liu, G. and Guo, J. (2011) Speech Enhancement Based on Modified a Priori SNR Estimation. Frontiers of Electrical & Electronic Engineering in China, 6, 542-546.
[Google Scholar] [CrossRef] 
[3] Xia, B. and Bao, C. (2014) Wiener Filtering Based Speech Enhancement with Weighted Denoising Auto-Encoder and Noise Classification. Speech Communication, 60, 13-29.
[Google Scholar] [CrossRef] 
[4] 沈锁金, 魏静, 高颖. 基于新型先验信噪比估计的语音增强算法的对比研究[J]. 中国集成电路, 2016, 210(11): 41-45.
[5] Ephraim, Y. and Malah, D. (1984) Speech Enhancement Using a Minimum Mean-Square Error Short-Time Spectral Amplitude Estimator. IEEE Transactions on Acoustics Speech & Signal Processing, 32, 1109-1121.
[Google Scholar] [CrossRef] 
[6] 沈锁金, 欧世峰，刘伟, 魏静. 基于先验信噪比估计语音增强算法的对比分析[J]. 烟台大学学报(自然科学与工程版), 2017, 30(2): 298-305.
[7] 欧世峰, 赵晓晖. 改进型先验信噪比估计语音增强算法[J]. 吉林大学学报: 工学版, 2009, 39(3): 787-791.
[8] Shen, S., Ou, S., Wei, J., et al. (2017) A Priori SNR Estimator Based on a Convex Combination of Two DD Approaches for Speech Enhancement. IEEE International Conference on Signal and Image Processing, 750-754.
[9] Boll, S.F. (1979) Suppression of Acoustic Noise in Speech Using Spectral Subtraction. IEEE Transactions on Acoustics Speech & Signal Processing, 27, 113-120.
[Google Scholar] [CrossRef] 
[10] Plapous, C., Marro, C. and Scalart, P. (2006) Improved Signal-to-Noise Ratio Estimation for Speech Enhancement. IEEE Transactions on Acoustics, Speech, and Signal Processing, 14, 2098-2108.
[11] 沈锁金. 语音增强技术中的先验信噪比估计算法研究[D]: [硕士学位论文]. 烟台: 烟台大学, 2017.
[12] Taal, C.H., Hendriks, R.C., Heusdens, R., et al. (2011) An Algorithm for Intelligibility Prediction of Time-Frequency Weighted Noisy Speech. IEEE Transactions on Audio Speech & Language Processing, 19, 2125-2136.
[Google Scholar] [CrossRef] 
[13] Abramson, A. and Cohen, I. (2010) Simultaneous Detection and Estimation Approach for Speech Enhancement. IEEE Transactions on Audio Speech & Language Processing, 15, 2348-2359.
[Google Scholar] [CrossRef]