提出了一种适用于抗噪声语音识别的方法,其特征提取过程基于最小方差无失真响应(M in im um varianced istortion less response,M VDR)谱估计方法,并对该特征进行频率弯折以提高其知觉分辨率,最后使用基于正则相关分析的谱变换补偿(C a...提出了一种适用于抗噪声语音识别的方法,其特征提取过程基于最小方差无失真响应(M in im um varianced istortion less response,M VDR)谱估计方法,并对该特征进行频率弯折以提高其知觉分辨率,最后使用基于正则相关分析的谱变换补偿(C anon ica l corre lation based on com pensation,CCBC)法对该特征进行自适应处理,从而提高了系统的鲁棒性。在展览馆噪声、人群噪声和汽车噪声下,与基于传统M e l倒谱系数(M FCC)特征的系统进行了对比实验,结果表明使用本文方法的语音识别系统的识别率得到了显著的提高。展开更多
Signal to noise ratio (SNR) and resolution are two important but contradictory characteristics used to evaluate the quality of seismic data. For relatively preserving SNR while enhancing resolution, the signal purit...Signal to noise ratio (SNR) and resolution are two important but contradictory characteristics used to evaluate the quality of seismic data. For relatively preserving SNR while enhancing resolution, the signal purity spectrum is introduced, estimated, and used to define the desired output amplitude spectrum after deconvolution. Since a real reflectivity series is blue rather than white, the effects of white reflectivity hypothesis on wavelets are experimentally analyzed and color compensation is applied after spectrum whitening. Experiments on real seismic data indicate that the cascade of the two processing stages can improve the ability of seismic data to delineate the geological details.展开更多
文摘提出了一种适用于抗噪声语音识别的方法,其特征提取过程基于最小方差无失真响应(M in im um varianced istortion less response,M VDR)谱估计方法,并对该特征进行频率弯折以提高其知觉分辨率,最后使用基于正则相关分析的谱变换补偿(C anon ica l corre lation based on com pensation,CCBC)法对该特征进行自适应处理,从而提高了系统的鲁棒性。在展览馆噪声、人群噪声和汽车噪声下,与基于传统M e l倒谱系数(M FCC)特征的系统进行了对比实验,结果表明使用本文方法的语音识别系统的识别率得到了显著的提高。
基金supported by the National Natural Science Foundation of China(Grant No.41174117)PetroChina Innovation Foundation(Grant No.2010D-5006-0301)
文摘Signal to noise ratio (SNR) and resolution are two important but contradictory characteristics used to evaluate the quality of seismic data. For relatively preserving SNR while enhancing resolution, the signal purity spectrum is introduced, estimated, and used to define the desired output amplitude spectrum after deconvolution. Since a real reflectivity series is blue rather than white, the effects of white reflectivity hypothesis on wavelets are experimentally analyzed and color compensation is applied after spectrum whitening. Experiments on real seismic data indicate that the cascade of the two processing stages can improve the ability of seismic data to delineate the geological details.