基于區(qū)域增長校正的頻域盲源分離排序算法
doi: 10.11999/JEIT180386
-
重慶郵電大學(xué) 信號與信息處理重慶市重點實驗室 ??重慶 ??400065
Frequency Domain Blind Source Separation Permutation Algorithm Based on Regional Growth Correction
-
Chongqing Key Laboratory of Signal and Information Processing, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
-
摘要:
卷積盲源分離可以在頻域得到有效解決，但頻域盲源分離必須解決排序模糊問題。該文提出一種基于區(qū)域增長校正的頻域盲源分離排序算法。首先對卷積混合信號短時傅里葉變換，在頻域的各個頻點處建立瞬時模型進行獨立分量分析，在此基礎(chǔ)上使用分離信號功率比的相關(guān)性，對所有頻點進行逐點排序置換。其次根據(jù)閾值將排序后的結(jié)果劃分為若干個小區(qū)域。最后按區(qū)域增長方式進行區(qū)域置換與合并，最終得到正確的分離信號。區(qū)域增長校正可最大限度地減少頻點排序錯誤擴散現(xiàn)象，從而改善分離效果。在模擬和真實環(huán)境中分別進行語音盲源分離實驗，結(jié)果表明所提算法的有效性。
-
關(guān)鍵詞:
- 卷積盲源分離 /
- 頻域排序 /
- 區(qū)域增長 /
- 功率比相關(guān)
Abstract:The convolutive blind source separation can be effectively solved in frequency domain, but blind source separation in frequency domain must solve the problem of ranking ambiguity. A frequency-domain blind source separation sorting algorithm is proposed based on regional growth correction. First, the convolutional mixed signal short-time Fourier transform is used to establish an instantaneous model at each frequency point in the frequency domain for independent component analysis. Based on this, the correlation of the power ratio of the separated signal is used to sort all frequency points one by one replacement. Second, according to the threshold, the sorted result is divided into several small areas. Finally. regional replacement and merging is performed according to the regional growth method, and the correct separation signal is finally obtained. Regional growth correction minimizes the mis-proliferation of frequency sorting and improves separation results. The speech blind source separation experiments are performed in the simulated and real environments respectively. The results show the effectiveness of the proposed algorithm.
-
表 1 算法性能對比(dB)
性能指標(biāo) 分離信號1 分離信號2 Murata算法 本文算法 Murata算法 本文算法 SIR 6.4071 15.8474 8.5336 18.7533 SDR 5.3447 7.6937 5.6878 9.4011 SAR 4.8792 8.8522 8.0340 10.1978 
下載: 導(dǎo)出CSV
表 2 本文算法復(fù)雜度
各算法塊 計算量 功率比計算 $\left( {L/2} \right) \left( {{N^2}B + NB} \right)$ 逐點排序 $\left( {L/2} \right) {N^2}B$ 區(qū)域排序 $R {N^2}B$
下載: 導(dǎo)出CSV
-
張?zhí)祢U, 馬寶澤, 強幸子, 等. 一種引入自適應(yīng)動量項的變步長混沌信號盲分離算法[J]. 電子與信息學(xué)報, 2017, 39(4): 908–914. doi: 10.11999/JEIT160578ZHANG Tianqi, MA Baoze, QIANG Xingzi, et al. Variable-step blind source separation algorithm with adaptive momentum item for chaotic signals[J]. Journal of Electronics &Information Technology, 2017, 39(4): 908–914. doi: 10.11999/JEIT160578 歐世峰, 耿超, 高穎. 動量項盲源分離算法及其性能優(yōu)化策略[J]. 電子學(xué)報, 2014, 42(1): 42–48. doi: 10.3969/j.issn.0372-21122014.01.007OU Shifeng, GENG Chao, and GAO Ying. Momentum term based blind source separation algorithm and its performance modified strategies[J]. Acta Electronica Sinica, 2014, 42(1): 42–48. doi: 10.3969/j.issn.0372-21122014.01.007 徐成發(fā), 郝宇星, 陸潞, 等. 基于互相關(guān)的快速角度估計算法[J]. 電子與信息學(xué)報, 2016, 38(6): 1446–1451. doi: 10.11999/JEIT151021XU Chengfa, HAO Yuxing, LU Lu, et al. Fast angle estimation algorithm based on cross-correlation[J]. Journal of Electronics &Information Technology, 2016, 38(6): 1446–1451. doi: 10.11999/JEIT151021 胡可, 汪增福. 一種基于時頻分析的語音卷積信號盲分離算法[J]. 電子學(xué)報, 2006, 34(7): 1246–1254. doi: 10.3321/j.issn:0372-2112.2006.07.014HU Ke and WANG Zengfu. A time-frequency analysis based blind source deconvolution method[J]. Acta Electronica Sinica, 2006, 34(7): 1246–1254. doi: 10.3321/j.issn:0372-2112.2006.07.014 YOSHIOKA T, NAKATANI T, MIYOSHI M, et al. Blind separation and dereverberation of speech mixtures by joint optimization[J]. IEEE Transactions on Audio, Speech and Language Processing, 2011, 19(1): 69–84. doi: 10.1109/TASL.2010.2045183 歐旭東, 張?zhí)祢U, 閆振華, 等. 基于多頻段能量相關(guān)排序的語音卷積混合盲源分離[J]. 計算機應(yīng)用研究, 2016, 33(5): 1481–1485. doi: 10.3969/j.issn.1001-3695.2016.05.046OU Xudong, ZHANG Tianqi, YAN Zhenhua, et al. Blind convolution speech separation based on multi-band ordering[J]. Application Research of Computers, 2016, 33(5): 1481–1485. doi: 10.3969/j.issn.1001-3695.2016.05.046 PARRA L and SPENCE C. Convolutive blind separation of non-stationary sources[J]. IEEE Transactions on Speech Audio Process, 2000, 8(3): 320–327. doi: 10.1109/89.841214 IKRAM M and MORGAN D. Permutation inconsistency in blind speech separation: Investigation and solutions[J]. IEEE Transactions on Speech & Audio Process, 2005, 13(1): 1–13. 張華, 馮大政, 龐繼勇. 基于二階統(tǒng)計量的語音信號時域卷積盲分離算法[J]. 電子與信息學(xué)報, 2010, 32(5): 1083–1086. doi: 10.3724/SP.J.1146.2009.00574ZHANG Hua, FENG Dazheng, and PANG Jiyong. A time-domain blind convolutive separation algorithm for speech signals based on second-order statistic[J]. Journal of Electronics &Information Technology, 2010, 32(5): 1083–1086. doi: 10.3724/SP.J.1146.2009.00574 SAWADA H, MUKAI R, ARAKI S, et al. A robust and precise method for solving the permutation problem of frequency domain blind source separation[J]. IEEE Transactions on Speech Audio Process, 2004, 12(5): 530–538. doi: 10.1109/TSA.2004.832994 MIETTINEN J, NORDHAUSEN K, and TASKINEN S. Blind source separation based on joint diagonalization in R: The packages JADE and BSSasymp[J]. Journal of Statistical Software, 2017, 76(2): 1–31. doi: 10.18637/jss.v076.i02 MURATA N, IKEDA S, and ZIEHE A. An approach to blind source separation based on temporal structure of speech signals[J]. Neurocomputing, 2001, 41(1): 1–24. 薄祥雷, 何怡剛, 尹柏強, 等. 基于影響因子的頻域盲源分離排序算法[J]. 電子學(xué)報, 2013, 42(2): 360–365. doi: 10.3969/j.issn.0372-2112.2014.02.023BO Xianglei, HE Yigang, YIN Baiqiang, et al. Algorithm to eliminate permutation of frequency domain blind source separation based on influence factor[J]. Acta Electronica Sinica, 2013, 42(2): 360–365. doi: 10.3969/j.issn.0372-2112.2014.02.023 SAITO S, OISHI K, and FURUKAWA T. Convolutive blind source separation using an iterative least-squares algorithm for non-orthogonal approximate joint diagonalization[J]. IEEE/ACM Transactions on Audio Speech & Language Processing, 2015, 23(12): 2434–2448. -
計量
- 文章訪問數(shù): 1579
- HTML全文瀏覽量: 492
- PDF下載量: 56
- 被引次數(shù): 0
下載:
