Voice conversion is a technique that aims to transform the individuality of source speech so as to mimic that of target speech while keeping the message unaltered, where the Gaussian mixture model based methods are mo...
详细信息
ISBN:
(纸本)9781509018918
Voice conversion is a technique that aims to transform the individuality of source speech so as to mimic that of target speech while keeping the message unaltered, where the Gaussian mixture model based methods are most commonly used. However, these methods suffer from over-smoothing and over-fitting problems. In our previous work, we proposed to use Gaussian processes to alleviate over-fitting. Despite its effectiveness, this method will inevitably lead to over-smoothing due to choosing the mean of predictive distribution of Gaussian processes as optimal estimation. Thus, in this paper we focus on addressing the over-smoothing problem by post-filtering the outputs of the standard Gaussian processes, resulting in more dynamics in the converted feature parameters. Experiments have confirmed the validity of the proposed method both objectively and subjectively.
In recent years, both of financial technologies and its services for mobile devices have been increased sharply. However, according to the case studies of mobile security, security threats have been increased too. Thu...
详细信息
ISBN:
(纸本)9781509009855
In recent years, both of financial technologies and its services for mobile devices have been increased sharply. However, according to the case studies of mobile security, security threats have been increased too. Thus, for sustaining their services safe, security department of companies or bank institutes continuously research and develop to upgrade their service solutions. However, recent researches of security threats reported that various new types of social engineering attack were frequently observed in mobile service. In this paper, we compare several user authentication schemes against social engineering attacks, especially shoulder surfing attack. In the experimental result, each one of existing schemes have a slightly different weaknesses. According to the result of comparative analysis, we suggest some of considerations for developing anti-shoulder surfing security solution.
Flexible micro-perforated panel has unique advantages in noise reduction due to its good flexibility compared with traditional rigid micro-perforated panel. In this paper, flexible micro-perforated panel was prepared ...
Flexible micro-perforated panel has unique advantages in noise reduction due to its good flexibility compared with traditional rigid micro-perforated panel. In this paper, flexible micro-perforated panel was prepared by computer numerical control (CNC) milling machine. Three kinds of plastics including polyvinylchloride (PVC), polyethylene terephthalate (PET), and polyimide (PI) were taken as the matrix materials to prepare flexible micro-perforated panel. It has been found that flexible micro-perforated panel made of PET possessing good porosity and proper density, elastic modulus and poisson ratio exhibited the best acoustic absorption properties. The effects of various structural parameters including perforation diameter, perforation ratio, thickness and air gap have also been investigated, which would be helpful to the optimization of acoustic absorption properties.
In this study, we focus on the discontinuity of samples distribution in the feature space due to the existence of different stress levels. Some parameters from stress samples, calculated using a mathematical model, ar...
详细信息
ISBN:
(纸本)9781479987467
In this study, we focus on the discontinuity of samples distribution in the feature space due to the existence of different stress levels. Some parameters from stress samples, calculated using a mathematical model, are clustered to different virtual classes according to degrees of stress. Based on different stress level cluster analysis is proposed within stress class. Experimental results illustrate that the performance for differentiating different speaking styles has been improved.
In this paper, a computationally efficient voice conversion system has been designed in order to improve the performance in resource-limited scenarios. First, mixtures of Gaussians (MoGs) at fixed locations of Mel fre...
详细信息
ISBN:
(纸本)9781479987467
In this paper, a computationally efficient voice conversion system has been designed in order to improve the performance in resource-limited scenarios. First, mixtures of Gaussians (MoGs) at fixed locations of Mel frequencies have been used to represent the spectrum of STRAIGHT compactly. Second, the key conditional distributions for prediction are approximated by building histograms of aligned features empirically. Experiments have confirmed that our proposed method can obtain fairly good results compared to the traditional method without huge computational costs.
This research aims to recognize the defect of concrete materials using an ultrasonic computed tomography imaging technique. Filtered Backprojection method was used to reconstruct concrete images in this paper. Ultraso...
详细信息
This research aims to recognize the defect of concrete materials using an ultrasonic computed tomography imaging technique. Filtered Backprojection method was used to reconstruct concrete images in this paper. Ultrasonic time of flight data was measured to reconstruct computer tomography images. 306 data paths were obtained in total by manual scanning for one computer tomography image. We examined the effect of the interpolation data as the density of time of flight data has a considerable effect on image quality. The feasibility of concrete reconstruction system and time of flight data interpolation were examined in detail using numerical and concrete phantoms.
暂无评论