基于低秩張量補(bǔ)全的多聲道音頻信號恢復(fù)方法
doi: 10.11999/JEIT150589
-
2.
(北京理工大學(xué)信息與電子學(xué)院 北京 100081) ②(內(nèi)蒙古科技大學(xué)信息工程學(xué)院 包頭 014010)
基金項(xiàng)目:
國家自然科學(xué)基金(61473041)，內(nèi)蒙古高?？蒲许?xiàng)目(NJZY13139)
Low Rank Tensor Completion for Recovering Missing Data in Multi-channel Audio Signal
-
2.
(School of Information and Electronics, Beijing Institute of Technology, Beijing 100081, China)
Funds:
The National Natural Science Foundation of China (61473041), Scientific Research Project in Colleges and Universities of Inner Mongolia (NJZY13139)
-
摘要: 多聲道音頻信號在采集、壓縮、傳輸過程中可能造成音頻數(shù)據(jù)丟失，為了確保給聽眾帶來更真實(shí)的聽覺感受，該文提出一種基于低秩張量補(bǔ)全的音頻丟失數(shù)據(jù)恢復(fù)方法。首先，把多聲道音頻信號表示為一個張量；其次，把張量補(bǔ)全作為一個凸優(yōu)化問題建模，利用松弛技術(shù)和變量分離技術(shù)得到閉合的增強(qiáng)拉格朗日函數(shù)；最后，通過交替迭代方法求解得到恢復(fù)的音頻張量。在不同數(shù)據(jù)丟失率的實(shí)驗(yàn)中，通過與線性預(yù)測、加權(quán)優(yōu)化的CANDECOMP /PARAFAC分解方法進(jìn)行對比分析，表明利用張量補(bǔ)全方法具有更高的音頻信號恢復(fù)精度，隱藏參考和基準(zhǔn)的多激勵測試結(jié)果也顯示低秩張量補(bǔ)全方法能夠有效地恢復(fù)多聲道音頻的丟失數(shù)據(jù)，從而獲得更好的聽覺效果。
-
關(guān)鍵詞:
- 音頻信號恢復(fù) /
- 張量補(bǔ)全 /
- 跡范數(shù) /
- 凸優(yōu)化
Abstract: The data maybe miss due to problems in the acquisition, compression or transmission process of multi- channel audio signal. In order to take audiences real auditory sense, an approach of signal recovery based on low rank tensor completion is proposed. First, multi-channel audio signal is represented as a signal tensor. Second, tensor completion is formulated as a convex optimization problem. A closed form for augmented Lagrangian function is obtained via relaxation technique and separation of variables technique. At last, the audio tensor is recovered by alternating iteration. In experiments of varying number of missing entries, the comparisons show that the proposed method is more accurate than linear prediction and CANDECOMP/PARAFAC weighted optimization. The results of multiple stimuli with hidden reference and anchor indicate that low rank tensor completion method is validated for multi-channel audio signal recovery. The better auditory effects are obtained by recovered audio.-
Key words:
- Audio signal recovery /
- Tensor completion /
- Trace norm /
- Convex optimization
-
王磊, 周樂囡, 姬紅兵, 等. 一種面向信號分類的匹配追蹤新方法[J]. 電子與信息學(xué)報, 2014, 36(6): 1299-1306. doi: 10.3724/SP.J.1146.2013.00942. WANG Lei, ZHOU Lenan, JI Hongbing, et al. A new matching pursuit algorithm for signal classification[J]. Journal of Electronics Information Technology, 2014, 36(6): 1299-1306. doi: 10.3724/SP.J.1146.2013.00942. VASEGHI S and FRANYLING C. Restoration of old gramophone recordings[J]. AES Journal of the Audio Engineering Society, 1992, 40(10): 791-801. 高悅, 陳硯圃, 閔剛, 等. 基于線性預(yù)測分析和差分變換的語音信號壓縮感知[J]. 電子與信息學(xué)報, 2012, 34(6): 1408-1413. doi: 10.3724/SP.J.1146.2011.01001. GAO Yue, CHEN Yanpu, MIN Gang, et al. Compressed sensing of speech signals based on linear prediction coefficients and difference transformation[J]. Journal of Electronics Information Technology, 2012, 34(6): 1408-1413. doi: 10.3724/SP.J.1146.2011.01001. COCCHI G and UNCINI A. Subbands audio signal recovering using neural nonlinear prediction[C]. Proceedings of the 2001 International Conference on Acoustics, Speech and Signal Processing (ICASSP), Salt Lake City, UT, USA, 2001: 1289-1292. 朱墨, 吳國清, 郭新毅. 基于盲解卷積的水聲信號恢復(fù)技術(shù)[J].應(yīng)用聲學(xué), 2011, 30(3): 177-186. doi:10.3969/j.issn. 1000- 310X.2011.03.003. ZHU Mo, WU Guoqing, and GUO Xinyi. An underwater signal recovery technique based on blind deconvolution[J]. Journal of Applied Acoustics, 2011, 30(3): 177-186. doi: 10.3969/j.issn.1000-310X.2011.03.003. ACAR E, DUNLAVY D M, KOLDA T G, et al. Scalable tensor factorizations with missing data[C]. Proceedings of the 10th SIAM International Conference on Data Mining, Columbus, OH, United States, 2010: 701-712. ZHAO Qibin, ZHANG Liqing, and CICHOCKI A. Bayesian CP factorization of incomplete tensors with automatic rank determination[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 37(9): 1751-1763. doi: 10.1109/TPAMI.2015.2392756. TAN Huachun, WU Yuankai, FENG Guangdong, et al. A new traffic prediction method based on dynamic tensor completion[J]. Procedia-Social and Behavioral Sciences, 2013, 96(8): 2431-2442. doi: 10.1016/j.sbspro.2013.08.272. LIU Yuanyuan and SHANG Fanhua. An efficient matrix factorization method for tensor completion[J]. IEEE Signal Processing Letters, 2013, 20(4): 307-310. doi:10.1109/LSP. 2013.2245416. 劉園園. 快速低秩矩陣與張量恢復(fù)的算法研究[D]. [博士論文] ,西安電子科技大學(xué), 2013. doi: 10.7666/d.D363665. LIU Yuanyuan. Algorithm research of fast low-rank matrix and tensor recovery[D]. [Ph.D. dissertation], Xidian University, 2013. doi: 10.7666/d.D363665. 樊勁宇, 顧紅, 蘇衛(wèi)民, 等. 基于張量分解的互質(zhì)陣MIMO 雷達(dá)目標(biāo)多參數(shù)估計方法[J]. 電子與信息學(xué)報, 2015, 37(4): 933-938. doi: 10.11999/JEIT140826. FAN Jinyu, GU Hong, SU Weimin, et al. Co-prime MIMO radar multi-parameter estimation based on tensor decomposition[J]. Journal of Electronics Information Technology, 2015, 37(4): 933-938. doi: 10.11999/JEIT140826. CICHOCKI A, ZDUNEK R, PHAN A H, et al. Nonnegative matrix and tensor factorizations[M]. Chichester, WS: John Wiley Sons, 2009: 28-31. LERMAN G and ZHANG T. Robust recovery of multiple subspaces by geometric lp minimization[J]. Annals of Statistics, 2011, 39(5): 2686-2715. doi: 10.1214/11-AOS914. CHEN Y, HSU C, and LIAO H M. Simultaneous tensor decomposition and completion using factor priors[J]. IEEE Transactions on Software Engineering, 2014, 36(3): 577-591. doi: 10.1109/TPAMI.2013.164. RECHT B, FAZEL M, and PARRILO P. Guaranteed minimum-rank solutions of linear matrix equations via nuclear norm minimization[J]. SIAM Review, 2010, 52(3): 471-501. LIU Ji, MUSIALSKI P, WONKA P, et al. Tensor completion for estimating missing values in visual data[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(1): 208-2121. doi: 10.1109/TPAMI.20125.39. GANDY S, RECHT B, and YAMADA I. Tensor completion and low-n-rank tensor recovery via convex optimization[J]. Inverse Problems, 2011, 27(2): 25010-25028. -
計量
- 文章訪問數(shù): 1673
- HTML全文瀏覽量: 205
- PDF下載量: 600
- 被引次數(shù): 0
下載:
