基于线性预测残差倒谱的多语音基音频率检测

本文提出了一种基于线性预测残差倒谱的多语音基音频率检测算法,该算法首先对混合语音信号进行线性预测分析,进而计算预测信号与原混合信号的残差,并对残差信号做倒谱变换,得到混合语音信号的线性预测残差倒谱;然后在该信号的残差倒谱中,结合图像处理的技术,利用语音信号基音倒频匹配法检测出多语音信号的基音频率;最后在基音标定的过程中,本文算法利用语音信号的连续特性,依据信号基音频率前后差距变化最小原则标记出各基音所属话者。实验结果表明,本文提出的算法在弱回声及无回声的情况下能快速有效地从单声道混合语音信号中检测出多语音基音信息。     

基于三阶累积量的语音信号的基音周期检测

本文提出了一种新的语音信号的基音周期检测方法,该方法根据语音信号的三阶累积量去确定语音信号的基音周期,能有效地排除白色或有色的高斯加性噪声所带来的干扰.与传统的基音周期估计的自相关函数法或平均幅度差函数法(AMDF)相比,该方法更精确、有效,具有更强的鲁棒性.     

一种改进的混合激励线性预测的基音周期估计算法

本文详细讨论了混合激励线性预测(MELP)的基音周期估计算法及其改进算法.该算法采用了分数基音周期、倍数检测等技术,保证了基音周期估计的精度.同时又采用了滑动窗的方法,使得对基音周期不规则的不平稳的语音段进行基音周期估计时的误差减小.本文最后给出了该算法的测试结果及优缺点.     

一种语音更改技术的研究与实现

基于时域基音同步叠加算法完成了对语音信号的更改。首先求出语音信号的基音周期,接着对语音信号进行基音标注,然后对基音周期进行更改,最后,将语音信号按照更改后的基音周期基于时域基音同步叠加算法进行语音合成。实验表明,语音更改方法可得到很好的效果。     

一种适于计算声场景分析的混叠语音基音检测方法

本文提出了一种在混叠语音信号中检测各自语音分量基音信息的方法.该方法采用小波变换作为基音检测模型中的滤波处理,并用广义自相关运算突出基音信息,用增强自相关累和消除冗余信息,并提出了用基音概率函数来预测并跟踪不同基音的变化以提高基音检测的准确性.本文提出的方法可应用于计算声场景分析中.实验结果表明,该方法对于混叠语音的基音检测是非常有效的.     

一种基音周期估计方法

基音是语音信号中一个极为重要的参数。基音周期的估计在语音编码、语音合成和语音识别中有着广泛的应用，本文介绍了一种极为重要的SIFT(Simplified inverse filter tracking简化逆滤波跟踪)基音周期估计算法，并对该算法进行了仿真，仿真结果表明，SIFT基因周期估计算法具有较好的估计性能。     

基于小波变换的语音基音周期检测

语音信号是一种非平稳信号,基音周期是语音信号最重要的参数之一,传统的基音检测方法存在一些缺陷.小波变换鲁棒性强、能很好地反映信号的时频特性,非常适合处理非平稳信号.为准确提取基音频率,提出了一种基于小波变换的基音周期检测方法.检测前在小波域上用Teager能量算子分离出语音信号的浊音段,然后对浊音段采用空域相关函数降噪...     

基于小波变换的自相关基音检测方法

基音周期是语音信号最重要的参数之一,其描述了语音激励源的一个重要特征。在噪声环境下,基音检测的准确率必然受到影响。文章提出基于小波变换的自相关(ACF)基音检测。通过实验仿真表明,该方法可有效地在信噪比较低的条件下提取语音的基音周期。     

基于SHS的重叠语音基音分离检测方法

运用SHS(分谐波累加)法可准确地提取单个语音的基音周期,本文将该方法推广运用到重叠语音基音的分离检测,并提出了相应的检测方法.实验证明,本文提出的方法能有效地分离并提取两重叠语音的基音,即使对于基音周期相差较近的两个重叠语音也能取得较好的结果.     

基于CEP和LPC谱提取语音信号基音周期的方法

在语音信号分析中,只有分析出可表示语音信号本质特征的参数,才有可能利用这些参数进行高效的语音通信、语音合成和语音识别等处理.因此对语音信号采用CEP和LPC谱提取语音信号基音周期的异同进行了研究,并采用Matlab实现了仿真分析.从中可以看出,LPC谱估计基音周期的算法运算量较大,而CEP谱算法更直观,且在少部分情况下基音峰会变得更突出一些,CEP谱具有更加广阔的应用前景.     

一种改进的基于DCT的语音增强算法

提出一种改进的基于离散余弦变换的语音增强算法。在信噪比较低时,传统的基于离散余弦变换的语音增强算法效果较好,能较大幅度地提高信号的信噪比;而当信噪比高时,利用这种方法会滤掉一些有用的信号成份。新算法首先计算出所有高阶离散余弦变换系数对应的时域信号中语音信号出现的可能性大小,然后根据某个阈值计算是否在估计噪声信号绝对值的均方差时保留该系数。实验结果表明在含噪语音信号的信噪比高于10dB时,新算法较传统的基于离散余弦变换的算法具有较好的性能。     

语音重构的DCT域加速Landweber迭代硬阈值算法

重构信号的最基本理论依据是该信号在某个变换域是稀疏的或近似稀疏的。基于语音信号在DCT域的近似稀疏性,可以采用压缩感知(Compressed Sensing, CS)理论对其进行重构。压缩感知理论中的迭代硬阈值(Iterative hard thresholding, IHT)算法以其较好的性能被广泛用来重构信号,但其收敛速度比较慢,如何提高收敛速度,一直是迭代硬阈值算法研究的重点之一。针对压缩感知理论中的IHT算法收敛速度相当慢的问题,提出了语音重构的DCT域加速Landweber迭代硬阈值(Accelerated Landweber iterative hard thresholding, ALIHT)算法。该算法对原始语音信号做DCT变换,然后在DCT域将每一步Landweber迭代分解为矩阵计算和求解两步,通过修改其中的矩阵计算部分实现Landweber迭代加速,最后通过迭代硬阈值对信号做阈值处理。实验结果表明,加速Landweber迭代硬阈值算法加快了收敛速度、减少了计算量。      

基于拉普拉斯分布模型的快速DCT算法

为提高编码效率,通过分析残差系数在空域和DCT域均符合拉普拉斯分布后,提出一种快速DCT算法.该算法能够在DCT之前对每个量化DCT系数进行零值预判而节省DCT计算.通过头肩序列的实验表明新算法在不降低图像质量的条件下,其整体运算复杂度优于常规算法.     

基于离散余弦变换的相位编码信号识别

针对低信噪比下相位差分法无法识别相位编码信号问题,提出了基于离散余变换(DCT)的相位编码信号识别方法。根据常规信号与相位编码信号的频谱差异,该方法采用平方运算对信号进行变换,利用离散余弦变换对平方后的信号频谱进行分析。在仿真条件相同的情况下,相比相位差分法,该方法识别率较高,抗噪性能也有所提高。同时,该方法具有运算量小和提取特征参量稳定的特点。     

基于Gamma语音模型的语音增强算法

提出了一种新的基于Gamma语音模型的语音增强算法。首先,在假定语音和噪声的短时DCT系数分别服从Gamma和Gaussian分布的基础上,推导了最小均方误差意义下的语音信号短时DCT系数估计;然后,根据语音存在概率估计,提出了语音信号短时DCT系数估计的修正因子。在增强算法中,提出了基于Gamma语音模型的改进最小统计量控制递归平均(IMCRA)噪声估计算法。仿真结果表明,该算法不仅在噪声抑制性能方面优于近两年国际上提出的几种基于Gaussian语音模型的语音增强算法,而且在增强语音质量方面也具有更好的性能。     

A Motion Picture Coding Algorithm Using Adaptive DCT Encoding Based on Coefficient Power Distribution Classification

A motion picture coding algorithm using motion-compensated interframe prediction and the adaptive discrete cosine transform (DCT) encoding technique is proposed. High coding efficiency is obtained by the adaptive DCT encoding technique in which encoding parameters are fitted to widely varying characteristics of the interframe differential signal. Segmented DCT subblocks of interframe prediction error are classified into categories based on their coefficient power distribution characteristics. The adaptation gain results from using a suitable variable word length code set designated by the above classification for encoding each quantization index of DCT coefficients. In addition, a new coding parameter control method is introduced based on the information rate estimation of the current frame. This classification promotes high stability because good estimation accuracy of bits consumption for each DCT subblock is obtained by utilizing the category indexes. Simulation results show that the proposed algorithm has enough coding efficiency to transmit videoconferencing motion pictures through a 384 kbit/s channel.     

The New Peak-to-Average Power Reduction Algorithm in the OFDM System

In this paper, we proposed a new peak-to-average power reduction (PAPR) algorithm of orthogonal frequency division multiplexing (OFDM) system using block coding scheme and discrete cosine transform (DCT). We are using DCT to concentrate the energy of the original signal into a few coefficients. After the DCT data were fed into the IDFT, the output of signal of OFDM appeared to have uniform distribution. With the newly proposed schemes, that we founded those three important properties, the first property is the PAPR used be reduced by 9.4419 dB for BPSK mapper. The second property is the OFDM signals have capability of noise immunity and of error correction. And the third property is the effect of PAPR reduced can be implement by cascaded different method.     

Optimal coding of vectorcardiographic sequences using spatial prediction.

This paper discusses principles, implementation details, and advantages of sequence coding algorithm applied to the compression of vectocardiograms (VCG). The main novelty of the proposed method is the automatic management of distortion distribution controlled by the local signal contents in both technical and medical aspects. As in clinical practice, the VCG loops representing P, QRS, and T waves in the three-dimensional (3-D) space are considered here as three simultaneous sequences of objects. Because of the similarity of neighboring loops, encoding the values of prediction error significantly reduces the data set volume. The residual values are de-correlated with the discrete cosine transform (DCT) and truncated at certain energy threshold. The presented method is based on the irregular temporal distribution of medical data in the signal and takes advantage of variable sampling frequency for automatically detected VCG loops. The features of the proposed algorithm are confirmed by the results of the numerical experiment carried out for a wide range of real records. The average data reduction ratio reaches a value of 8.15 while the percent root-mean-square difference (PRD) distortion ratio for the most important sections of signal does not exceed 1.1%.     

On an aperiodic stochastic resonance signal processor and its application in digital watermarking

In this paper, an aperiodic stochastic resonance (ASR) signal processor for communication systems based on a bistable dynamic mechanism is proposed for detecting base-band binary pulse amplitude modulation (PAM) signals in communication systems. All parameters in the processor can be adjusted when needed. The adjustment mechanism is explained from the perspective of the conventional noise-induced nonlinear signal processing. To demonstrate this processor's usability, a digital image-watermarking algorithm in the discrete cosine transform (DCT) domain is implemented. In this algorithm, the watermark and the DCT alternating current (ac) coefficients of the image are viewed as the input signal and the channel noise, respectively. In conventional watermarking systems, it is difficult to explain why the detection bit error ratio (BER) of a watermarking system suffering from certain attacks is lower than that of the system not suffering from any attack. In the new watermarking algorithm, this phenomenon is systematically analyzed. It is shown that the DCT ac coefficients of an image as well as the noise imposed by the attacks can cooperate within the bistable system to improve the performance of the watermark detection.     

Computation reduction for standard-based video encoders based on the energy preservation property of DCT

Based on the energy preservation property of DCT, an optimization technique for motion estimation (ME), DCT, and quantization for standard-based video encoders is developed. First, a stopping criterion for ME is proposed to reduce the number of checking points in finding the motion vectors, and save the computations. The advantage of introducing such a stopping criterion lies in its adaptability to the quantization parameter and applicability to various fast ME algorithms. Then, the DCT and quantization are jointly optimized by tracing the remaining signal energy and removing unnecessary calculations in the process of DCT and quantization. A pruned 2-D DCT based on Huang's fast DCT algorithm is presented to demonstrate the superiority of this algorithm to the full DCT and an existing all-zero block detection method. Although proved to be computationally efficient, the algorithms introduce no obvious quality loss.