A continuous differentiable wavelet threshold function for speech enhancement

Abstract

Enhanced speech based on the traditional wavelet threshold function had auditory oscillation distortion and the low signal-to-noise ratio (SNR). In order to solve these problems, a new continuous differentiable threshold function for speech enhancement was presented. Firstly, the function adopted narrow threshold areas, preserved the smaller signal speech, and improved the speech quality; secondly, based on the properties of the continuous differentiable and non-fixed deviation, each area function was attained gradually by using the method of mathematical derivation. It ensured that enhanced speech was continuous and smooth; it removed the auditory oscillation distortion; finally, combined with the Bark wavelet packets, it further improved human auditory perception. Experimental results show that the segmental SNR and PESQ (perceptual evaluation of speech quality) of the enhanced speech using this method increase effectively, compared with the existing speech enhancement algorithms based on wavelet threshold.

Keywords

continuous differentiable / wavelet threshold function / speech enhancement / Bark wavelet packet / non-fixed deviation / noise

References

[1] LuC T, TsengK F. A gain factor adapted by masking property and SNR variation for speech enhancement in colored-noise corruptions [J]. Computer Speech and Language, 2010, 24(4): 632-647

[2] KrishnamoorthyP, PrasannaS R M. Enhancement of noisy speech by temporal and spectral processing [J]. Speech Communication, 2011, 53(2): 154-174

[3] LuC T. Enhancement of single channel speech using perceptual-decision-directed approach [J]. Speech Communication, 2011, 53(4): 495-507

[4] GunawanT S, AmbikairajahE, EppsJ. Perceptual speech enhancement exploiting temporal masking properties of auditory system [J]. Speech Communication, 2010, 52(5): 381-393

[5] BarriA, DoomsA, SchelkensP. The near shift-invariance of the dual-tree complex wavelet transform tevisited [J]. Journal of Mathematical Analysis and Applications, 2012, 389(2): 1303-1314

[6] BarriA, DoomsA, SchelkensP. The near shift-invariance of the dual-tree complex wavelet transform revisited [J]. Journal of Mathematical Analysis and Applications, 2012, 389(2): 1303-1314

[7] WangC-l, ZhangC-l, ZhangP-tu. Denoising algorithm based on wavelet adapaive threshold [J]. Physics Procedia, 2012, 24A: 678-685

[8] DonohoD L. De-noising by soft-thresholding [J]. IEEE Transaction on Information Theroy, 1995, 41(3): 613-627

[9] GaoH Y, BruceA G. Waveshrink with firm shrinkage [J]. Statistica Sinica, 1997, 7(4): 855-874

[10] BacchelliS, PapiS. Filtered wavelet thresholding methods [C]. Proceedings of the 10th International Congress on Computational and Applied Mathematics, 2004BelgiumUniversity of Leuven39-52

[11] YinX-c, HanP, ZhangJ, ZhangF-q, WangN-ling. Application of wavelet transform in signal denoising [C]. Proceeding of the Second International Conference on Machine Learing and Cybernetics. Xi’an, 2003436-441

[12] BreimanL. Better subset regression using the nonnegative garrote [J]. Technometrics, 1995, 37(4): 373-384

[13] DuanQ-c, DengY-j, YingZ-gui. Research of algorithm for speech enhancement based on improved threshold function wavelet packet [J]. Communication Technology, 2009, 542: 86-88

[14] WangC-l, ZhangC-l, ZhangP-tu. Denoising algorithm based on wavelet adaptive threshold [J]. Physics Procedia, 2012, 24(A): 678-685

[15] DengY-juanSpeech denoising algorithm based on wavelet transform [D], 2009ChongqingChongqing University46-47

[16] LitvinY, CohenI. Sing-channel source separation of audio signals using Bark scale wavelet paceket decomposition [J]. Journal of Signal Processing Systems, 2011, 65(3): 339-350

[17] GaiG-h, QuL-sheng. Translation-invariant based adaptive threshold denoising for impact signal [J]. Chinese Journal of Mechanical Engineering, 2004, 17(4): 552-555

[18] LUN D P K, SHEN T W, HSUNG T C, HO D K C. Wavelet based speech presence probability estimator for speech enhancement [EB/OL]. [2012-06-28]. http://www.sciencedirect.com/science/article/pii/S1051200412001522#.

[19] TohidypourH R, SeyyedsalehiS A, BehboodH, RoshandelH. A new representation for speech frame recognition based on redundant wavelet filter banks [J]. Speech Communication, 2012, 54(2): 256-271

[20] JeonY Y, LeeS M. A speech enhancement algorithm to reduce noise and compensate for partial masking effect [J]. Journal of Central South University of Technology, 2011, 18(4): 1121-1127

[21] RixA W, BeerendsJ G, HollierM P, HekstraA P. Perceptual evaluation of speech quality(PESQ)—A new method for speech quality assessment of telephone networks and codes [C]. Proceedings of the International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2001749-752