A machine learning based speech enhancement method is proposed to improve the intelligibility of whispered speech. A binary mask estimated by a two-class support vector machine （SVM） classifier is used to synthesize...A machine learning based speech enhancement method is proposed to improve the intelligibility of whispered speech. A binary mask estimated by a two-class support vector machine （SVM） classifier is used to synthesize the enhanced whisper. A novel noise robust feature called Gammatone feature cosine coefficients （GFCCs） extracted by an auditory periphery model is derived and used for the binary mask estimation. The intelligibility performance of the proposed method is evaluated and compared with the traditional speech enhancement methods. Objective and subjective evaluation results indicate that the proposed method can effectively improve the intelligibility of whispered speech which is contaminated by noise. Compared with the power subtract algorithm and the log-MMSE algorithm, both of which do not improve the intelligibility in lower signal-to-noise ratio （SNR） environments, the proposed method has good performance in improving the intelligibility of noisy whisper. Additionally, the intelligibility of the enhanced whispered speech using the proposed method also outperforms that of the corresponding unprocessed noisy whispered speech.展开更多
The noises of remote sensing images, caused by imaging system and ground environment, negatively affect the accuracy and efficiency in extracting forest information from remote sensing images. The denoising is critica...The noises of remote sensing images, caused by imaging system and ground environment, negatively affect the accuracy and efficiency in extracting forest information from remote sensing images. The denoising is critical for image classifications for forest areas. The objective of this research is to assess the effectiveness of currently used spatial filtering methods for extracting with forest information related from Landsat 5 TM images. Five spatial filtering methods including low-pass filter, median filter, mean filter, sigma filter and enhanced self-adaptive filter were examined. A set of evaluation indices was designed to assess the ability of each denoising method for flatness, edge/boundary retention and enhancement. Based on the designed evaluation indices and visual assessment, it was found that sigma filter （D=1） and enhanced self-adaptive filter were the most effective denoising methods in classifying TM images for forest areas.展开更多
An improved speech absence probability estimation was proposed using environmental noise classification for speech enhancement.A relevant noise estimation approach,known as the speech presence uncertainty tracking met...An improved speech absence probability estimation was proposed using environmental noise classification for speech enhancement.A relevant noise estimation approach,known as the speech presence uncertainty tracking method,requires seeking the "a priori" probability of speech absence that is derived by applying microphone input signal and the noise signal based on the estimated value of the "a posteriori" signal-to-noise ratio(SNR).To overcome this problem,first,the optimal values in terms of the perceived speech quality of a variety of noise types are derived.Second,the estimated optimal values are assigned according to the determined noise type which is classified by a real-time noise classification algorithm based on the Gaussian mixture model(GMM).The proposed algorithm estimates the speech absence probability using a noise classification algorithm which is based on GMM to apply the optimal parameter of each noise type,unlike the conventional approach which uses a fixed threshold and smoothing parameter.The performance of the proposed method was evaluated by objective tests,such as the perceptual evaluation of speech quality(PESQ) and composite measure.Performance was then evaluated by a subjective test,namely,mean opinion scores(MOS) under various noise environments.The proposed method show better results than existing methods.展开更多
Froth image could strongly indicate the production status in mineral flotation process.Considering low contrast and sensitivity to noises and illumination of froth images in flotation cells,an improved image enhanceme...Froth image could strongly indicate the production status in mineral flotation process.Considering low contrast and sensitivity to noises and illumination of froth images in flotation cells,an improved image enhancement algorithm based on nonsubsampled contourlet transform (NSCT) and multiscale Retinex algorithm has been proposed.Nonsubsampled contourlet transform was firstly adopted to decompose the flotation froth images,ensure signals invariance and avoid the blurring edge.Secondly,a multiscale Retinex algorithm was used to enhance the lower frequency image and improve the brightness uniformity.Adaptive classification method based on Bayes atrophy threshold was proposed to eliminate noise,preserve strong edges,and enhance weak edges of band-pass sub-band images.Experiment shows that the proposed method could enhance the edge,contour,details and curb noise,and improve visual effects.Under-segmentation caused by noise and blurring edge has been solved,which lays a foundation for extracting foamy morphological flotation froth and analyzing grade.展开更多
Effects of noise on rate oscillations during CO oxidation on Pt（110） surface were investigated, both theoretically and numerically, by focusing on the interplay of internal noise （IN） due to stochasticity in react...Effects of noise on rate oscillations during CO oxidation on Pt（110） surface were investigated, both theoretically and numerically, by focusing on the interplay of internal noise （IN） due to stochasticity in reaction events, and external noise （EN） resulting from parameter perturbation. The surface is divided into cells of variable size which are assumed to be well mixed, and we consider the behavior inside a single cell. Attention is paid to parameter regions subthreshold of the deterministic Hopf bifurcation, where noise can induce stochastic oscillations, the signal-to-noise ratio （SNR） of which shows a maximum with the variation of noise intensity, known as coherent resonance （CR）. By stochastic normal theory, we show that IN and EN contribute in a weighted additive way to an effective noise that lead to CR, such that SNR shows a ridge shape in the D-1/√N plane, where D and 1/√N measures the strength of EN and IN, respectively. It is shown that for too large IN （EN）, CR behavior with EN （IN） no longer exists. Numerical simulations show good agreements with the theoretical results.展开更多
Defining the quantity K as the signal-to-noise ratio （SNR） and the normalized intensity fluctuation C（O） of a single-mode laser for bias signal modulation driven by color noises with colored correlation, the whole...Defining the quantity K as the signal-to-noise ratio （SNR） and the normalized intensity fluctuation C（O） of a single-mode laser for bias signal modulation driven by color noises with colored correlation, the whole output properties of the laser system is described by K. It is found that there is a maximum in the curves of K versus D, Q, and io. The optimization parameters are gained.展开更多
There has been a lot of research has been performed regarding diagnosing rolling element bearing faults using wavelet analysis, but almost all methods are not ideal for picking up fault signal characteristics under st...There has been a lot of research has been performed regarding diagnosing rolling element bearing faults using wavelet analysis, but almost all methods are not ideal for picking up fault signal characteristics under strong noise. Therefore, this paper proposes auto-correlation, cross-correlation and weighted average fault diagnosis methods based on wavelet transform (WT) de-noising which combine correlation analysis with WT for the first time. These three methods compute the auto-correlation, the cross-correlation and the weighted average of the measured vibration signals, then de-noise by thresholding and computing the auto-correlation of de-noised coefficients of WT and FFT of energy sequence. The simulation results indicate that all methods enhance the capabilities of fault diagnosis of rolling bearings and pick up the fault characteristics effectively.展开更多
Some two-microphone noise reduction techniques that work in the frequency domain exploit coherence function between two noisy signals. They have shown good results when noise signals on two sensors are uncorrelated, b...Some two-microphone noise reduction techniques that work in the frequency domain exploit coherence function between two noisy signals. They have shown good results when noise signals on two sensors are uncorrelated, but their per-formance decreases with correlated noises. Coherence based methods can be improved when the cross power spectral density (CPSD) of correlated noise signals is available. In this paper, we propose a new method for estimation of the CPSD of the noise, which is based on the minimum tracking technique. Despite the fact that the proposed estimator does not need to implement a voice activity detector (VAD), its performance is comparable to a CPSD estimator that uses an ideal VAD.展开更多
基金The National Natural Science Foundation of China (No.61231002,61273266,51075068,60872073,60975017, 61003131)the Ph.D.Programs Foundation of the Ministry of Education of China(No.20110092130004)+1 种基金the Science Foundation for Young Talents in the Educational Committee of Anhui Province(No. 2010SQRL018)the 211 Project of Anhui University(No.2009QN027B)
文摘A machine learning based speech enhancement method is proposed to improve the intelligibility of whispered speech. A binary mask estimated by a two-class support vector machine （SVM） classifier is used to synthesize the enhanced whisper. A novel noise robust feature called Gammatone feature cosine coefficients （GFCCs） extracted by an auditory periphery model is derived and used for the binary mask estimation. The intelligibility performance of the proposed method is evaluated and compared with the traditional speech enhancement methods. Objective and subjective evaluation results indicate that the proposed method can effectively improve the intelligibility of whispered speech which is contaminated by noise. Compared with the power subtract algorithm and the log-MMSE algorithm, both of which do not improve the intelligibility in lower signal-to-noise ratio （SNR） environments, the proposed method has good performance in improving the intelligibility of noisy whisper. Additionally, the intelligibility of the enhanced whispered speech using the proposed method also outperforms that of the corresponding unprocessed noisy whispered speech.
文摘The noises of remote sensing images, caused by imaging system and ground environment, negatively affect the accuracy and efficiency in extracting forest information from remote sensing images. The denoising is critical for image classifications for forest areas. The objective of this research is to assess the effectiveness of currently used spatial filtering methods for extracting with forest information related from Landsat 5 TM images. Five spatial filtering methods including low-pass filter, median filter, mean filter, sigma filter and enhanced self-adaptive filter were examined. A set of evaluation indices was designed to assess the ability of each denoising method for flatness, edge/boundary retention and enhancement. Based on the designed evaluation indices and visual assessment, it was found that sigma filter （D=1） and enhanced self-adaptive filter were the most effective denoising methods in classifying TM images for forest areas.
基金Project supported by an Inha University Research GrantProject(10031764) supported by the Strategic Technology Development Program of Ministry of Knowledge Economy,Korea
文摘An improved speech absence probability estimation was proposed using environmental noise classification for speech enhancement.A relevant noise estimation approach,known as the speech presence uncertainty tracking method,requires seeking the "a priori" probability of speech absence that is derived by applying microphone input signal and the noise signal based on the estimated value of the "a posteriori" signal-to-noise ratio(SNR).To overcome this problem,first,the optimal values in terms of the perceived speech quality of a variety of noise types are derived.Second,the estimated optimal values are assigned according to the determined noise type which is classified by a real-time noise classification algorithm based on the Gaussian mixture model(GMM).The proposed algorithm estimates the speech absence probability using a noise classification algorithm which is based on GMM to apply the optimal parameter of each noise type,unlike the conventional approach which uses a fixed threshold and smoothing parameter.The performance of the proposed method was evaluated by objective tests,such as the perceptual evaluation of speech quality(PESQ) and composite measure.Performance was then evaluated by a subjective test,namely,mean opinion scores(MOS) under various noise environments.The proposed method show better results than existing methods.
基金Project(61134006)supported by the National Natural Science Foundation of ChinaProject(2012BAF03B05)supported by the National Key Technology R&D Program of ChinaProject(11JJ6062)supported by Hunan Provincial Natural Science Foundation,China
文摘Froth image could strongly indicate the production status in mineral flotation process.Considering low contrast and sensitivity to noises and illumination of froth images in flotation cells,an improved image enhancement algorithm based on nonsubsampled contourlet transform (NSCT) and multiscale Retinex algorithm has been proposed.Nonsubsampled contourlet transform was firstly adopted to decompose the flotation froth images,ensure signals invariance and avoid the blurring edge.Secondly,a multiscale Retinex algorithm was used to enhance the lower frequency image and improve the brightness uniformity.Adaptive classification method based on Bayes atrophy threshold was proposed to eliminate noise,preserve strong edges,and enhance weak edges of band-pass sub-band images.Experiment shows that the proposed method could enhance the edge,contour,details and curb noise,and improve visual effects.Under-segmentation caused by noise and blurring edge has been solved,which lays a foundation for extracting foamy morphological flotation froth and analyzing grade.
文摘Effects of noise on rate oscillations during CO oxidation on Pt（110） surface were investigated, both theoretically and numerically, by focusing on the interplay of internal noise （IN） due to stochasticity in reaction events, and external noise （EN） resulting from parameter perturbation. The surface is divided into cells of variable size which are assumed to be well mixed, and we consider the behavior inside a single cell. Attention is paid to parameter regions subthreshold of the deterministic Hopf bifurcation, where noise can induce stochastic oscillations, the signal-to-noise ratio （SNR） of which shows a maximum with the variation of noise intensity, known as coherent resonance （CR）. By stochastic normal theory, we show that IN and EN contribute in a weighted additive way to an effective noise that lead to CR, such that SNR shows a ridge shape in the D-1/√N plane, where D and 1/√N measures the strength of EN and IN, respectively. It is shown that for too large IN （EN）, CR behavior with EN （IN） no longer exists. Numerical simulations show good agreements with the theoretical results.
基金Key Project of Education Bureau of Hubei Province of China
文摘Defining the quantity K as the signal-to-noise ratio （SNR） and the normalized intensity fluctuation C（O） of a single-mode laser for bias signal modulation driven by color noises with colored correlation, the whole output properties of the laser system is described by K. It is found that there is a maximum in the curves of K versus D, Q, and io. The optimization parameters are gained.
文摘There has been a lot of research has been performed regarding diagnosing rolling element bearing faults using wavelet analysis, but almost all methods are not ideal for picking up fault signal characteristics under strong noise. Therefore, this paper proposes auto-correlation, cross-correlation and weighted average fault diagnosis methods based on wavelet transform (WT) de-noising which combine correlation analysis with WT for the first time. These three methods compute the auto-correlation, the cross-correlation and the weighted average of the measured vibration signals, then de-noise by thresholding and computing the auto-correlation of de-noised coefficients of WT and FFT of energy sequence. The simulation results indicate that all methods enhance the capabilities of fault diagnosis of rolling bearings and pick up the fault characteristics effectively.
基金Project supported by the Iran Telecommunications Research Center (ITRC)
文摘Some two-microphone noise reduction techniques that work in the frequency domain exploit coherence function between two noisy signals. They have shown good results when noise signals on two sensors are uncorrelated, but their per-formance decreases with correlated noises. Coherence based methods can be improved when the cross power spectral density (CPSD) of correlated noise signals is available. In this paper, we propose a new method for estimation of the CPSD of the noise, which is based on the minimum tracking technique. Despite the fact that the proposed estimator does not need to implement a voice activity detector (VAD), its performance is comparable to a CPSD estimator that uses an ideal VAD.
