期刊文献+

基于AR-HMM在线能量调整的语音增强方法 被引量:6

Online Energy Adjustment Using AR-HMM for Speech Enhancement
在线阅读 下载PDF
导出
摘要 针对单通道语音增强技术对非平稳噪声的跟踪不准确、噪声抑制效果较差的问题,本文提出一种基于在线能量调整的语音增强方法.该方法以归一化临界带能量为特征,采用高斯混合模型对背景噪声进行分类,利用对应类型噪声的自回归隐马尔可夫模型(Auto-Regressive Hidden Markov Model,AR-HMM)和纯净语音的AR-HMM,在最小均方误差准则下估计语音和噪声的功率谱.考虑到非平稳环境中训练集和测试集的差异性,需在线调整语音模型和噪声模型中的能量,语音模型的能量调整采用迭代的期望最大化算法;噪声模型的能量调整则利用的是模型训练过程中的能量重估方法,并以最小值控制的递归平均算法确定噪声能量调整的初始值.在ITU-T G.160标准下对算法进行性能测试,测试结果表明,本文方法对非平稳噪声的跟踪效果较好,对噪声衰减量较大,收敛时间较短. Because the existing single channel speech enhancement technologies perform not w ell in the tracking and suppression of non-stationary noise,the speech enhancement method based on online energy adjustment is proposed.The normalized critical band energy parameters are employed as the feature in Gaussian mixture model( GM M) to distinguish the background noises.Based on the AR-HM M of clean speech and the noise of corresponding type,the pow er spectrums of speech and noise are estimated under minimum mean square error( M M SE) criteria.When the differences betw een the training data and test data are considered in the non-stationary noise environment,the online adjustment method for the speech and noise models is necessary.The scaling factor of speech energy is estimated w ith the iterative expectation maximization( EM) algorithm and the one of noise energy is estimated with the re-estimation approach similar to the training stage.And the initial scaling factor of noise energy is obtained by minima-controlled recursive averaging( M CRA) algorithm.The evaluation of the proposed method is performed under the standard of ITU-T G.160.The test results reveal that,comparing w ith the tw o reference methods,the proposed method performs w ell in non-stationary noise environments,including larger noise reduction and shorter convergence time.
出处 《电子学报》 EI CAS CSCD 北大核心 2014年第10期1991-1997,共7页 Acta Electronica Sinica
基金 国家自然科学基金(No.61072089) 北京市教育委员会科技发展计划重点项目(No.KZ201110005005)
关键词 语音增强 非平稳噪声 隐马尔可夫模型 高斯混合模型 speech enhancement non-stationary noise hidden M arkov model Gaussian mixture model
作者简介 何玉文 女,1988年生于北京,北京工业大学硕士研究生,主要研究方向为语音增强.E—mail:iamhyw@emails.Bjut.Edu.Cn 鲍长春 男,1965年生于内蒙古赤峰,博士,北京工业大学教授、博士生导师,IEEE高级会员,国际语音通信学会(ISCh)会员,亚太信号与信息处理学会(APSIAP)会员,中国电子学会理事,中国声学学会理事,信号处理学会委员.主要研究方向为语音与音频信号处理.E-mail:chchbao@bjut.Edu.Cn 夏丙寅男,1986年生于北京,北京工业大学博士生,主要研究方向为语音编码与增强.E—mail:xby-abc@emails.bjut.edu.Cn
  • 相关文献

参考文献12

  • 1Ephraim Y. A Bayesian estimation approach for speech en- hancement using hidden Markov models[J]. IEEE Transactious on Signal Processing, 1992,40(4) :725 - 735.
  • 2Ephraim Y. Gain-adapted hidden Markov models for recogni- tion of clean and noisy speech[J]. IEEE Transactions on Signal Processing, 1992,40(6) : 1303 - 1316.
  • 3Sameti H, Sheikhzadeh H, Deng L, Brennan R L. HMM-based strategies for enhancement of speech signals embedded in non- stationary noise [J]. IEEE Transactions on Speech and Audio Processing, 1998,6 (5) : 445 - 455.
  • 4Srinivasan S, Samuelsson J, Kleijn W B. Codebook-based Bayesian speech enhancement[A].IEEE International Confer- ence on Acoustics, Speech, and Signal Processing[C]. IEEE, 2005.1077 - 1080.
  • 5Zhao D Y, Kleijn W B. HMM-based gain modeling for en- hancement of speech in noise[J]. IEEE Transactions on Audio, Speech, and Language Processing, 2007,15 (3) : 882 - 892.
  • 6Zhao D Y, Kleijn W B, Ypma A, et al. Online noise estimation using stochastic-gain HMM for speech enhancement [J]. IEEE Transactions on Audio, Speech, and Language Processing, 2008,16(4) :835 - 846.
  • 7Srinivasan S, Samuelsson J, Kleijn W B. Codebook-based Bayesian speech enhancement for nonstationary environments [J]. IEEE Transactions on Audio, Speech, and Language Pro- cessing,2007,15(2) :441 - 452.
  • 8Varga A, Steeneken H J M. Assessment for automatic speech recognition: H. NOISEX-92: a database and an experiment to study the effect of additive noise on speech recognition systems [J]. Speech Communication, 1993,12 (3) :247 - 251.
  • 9Johnston J D. Transform coding of audio signals using percep- tual noise criteria[J]. IEEE Journal on Selected Areas in Com- munications, 1988,6(2) :314 - 323.
  • 10Ephraim Y. A minimum mean square error approach for speech enhancement[A]. International Conference on Acous- tics, Speech, and Signal Processing [ C ]. IEEE, 1990. 829 - 832.

同被引文献42

引证文献6

二级引证文献25

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部