改进相位谱补偿的语音增强方法

针对传统单通道语音增强方法中用带噪语音相位代替纯净语音相位重建时域信号，使得语音主观感知质量改善受限的情况，提出了一种改进相位谱补偿的语音增强算法。该算法提出了基于每帧语音输入信噪比的Sigmoid型相位谱补偿函数，能够根据噪声的变化来灵活地对带噪语音的相位谱进行补偿；结合改进DD的先验信噪比估计与语音存在概率算法（SPP）来估计噪声功率谱；在维纳滤波中结合新的语音存在概率噪声功率谱估计与相位谱补偿来提高语音的增强效果。相比传统相位谱补偿（PSC）算法而言，改进算法可以有效抑制音频信号中的各类噪声，同时增强语音信号感知质量，提升语音的可懂度。     

一种改进的维纳滤波语音增强算法

提出了一种改进的语音增强算法,该算法以基于先验信噪比估计的维纳滤波法为基础。首先通过计算无声段的统计平均得到初始噪声功率谱;其次,计算语音段间带噪语音功率谱,并平滑处理初始噪声功率谱和带噪语音功率谱,更新了噪声功率谱;最后,考虑了某频率点处噪声急剧增大的情况,通过计算带噪语音功率谱与噪声功率谱的比值,自适应地调整噪声功率谱。将该算法与其他基于短时谱估计的语音增强算法进行了对比实验,实验结果表明：该算法能有效地减少残留噪声和语音畸变,提高语音可懂度。     

谐波重构先验信噪比估计算法

先验信噪比单通道语音增强算法在信噪比较高时能有效地去除噪声,但在信噪比较低时语音高次谐波失真较为严重。针对此提出了一种基于谐波重构的先验信噪比估计算法,对增强后的信号加权求平方,进行功率谱的二次谱处理,以加强语音信号的周期性;再进行谐波重构,提升谐波分量。实验研究表明,该算法在低信噪比时能够有效地增强语音谐波分量,相对于先验信噪比估计的语音增强算法能够改善语音质量,减少语音失真。     

基于改进谐波恢复算法的语音增强方法

当信噪比较低时,语音信号的高次谐波部分会完全淹没在噪音中。针对该情况,提出一种基于改进谐波恢复算法的语音增强方法。对经过MMSE-LSA算法语音增强处理后的时域输出语音信号进行非线性处理,得到准周期冲激信号,并将其与原增强信号相乘,突出语音的谐波分量。实验结果表明,改进算法较好地解决了低信噪比时谐波失真的问题,相比传统谐波恢复算法能更好地改善语音高次谐波的质量。     

联合相位谱补偿优化的幅度谱掩蔽语音增强

针对现阶段大多数的监督性语音增强网络无法处理高度非结构化特性的相位谱的问题,提出一种基于多目标深度神经网络的相位谱估计的语音增强方法.将对非结构化相位的估计转化为对相位谱补偿(PSC)的估计,在训练阶段将PSC中的补偿因子作为训练目标之一,达到直接估计相位的目的;在此基础上,结合信噪比(SNR)改进PSC,提高PSC在...     

一种基于谱减的语音增强算法改进

为进一步降低噪声对采集语音的干扰,提出了一种新的谱减改进方法。采用阈值法对非平稳背景噪声信号进行估计,计算出先验信噪比,得到还原的纯净语音信号。用MATLAB实现了整个算法的仿真,并与传统谱减法结果相比较,仿真结果表明,该算法对非平稳噪声追踪性较好,在抑制背景噪声,减少音乐噪声前提下,提高了语音的可懂度,其计算复杂度也可以接受。     

改进的减谱法在语音增强中的应用

为消除语音信号中噪声,改善语音质量,本文提出一种改进的减谱法。首先根据每帧的功率谱动态调整谱减系数,然后通过维纳滤波法把各种噪声变换为类似白噪声的噪声,最后用原减谱法把该噪声去除。实验证明,该方法有较好的去噪效果。     

联合优化深度神经网络和约束维纳滤波的单通道语音增强方法

深度神经网络（Deep neural networks,DNNs）依靠其良好的特征提取能力,在语音增强任务中得到了广泛应用。为进一步提高深度神经网络的语音增强效果,提出一种将深度神经网络和约束维纳滤波联合训练优化的新型网络结构。该网络首先对带噪语音幅度谱进行训练并分别得到纯净语音和噪声的幅度谱估计,然后利用语音和噪声的幅度谱估计计算得到一个约束维纳增益函数,最后利用约束维纳增益函数从带噪语音幅度谱中估计出增强语音幅度谱作为网络的训练输出。对不同信噪比下的20种噪声进行的仿真实验表明,无论噪声类型是否在网络的训练集中出现,本文方法都能够在有效去除噪声的同时保持较小的语音失真,增强效果明显优于DNN及NMF增强方法。     

利用ERB尺度划分对补偿相位谱语音增强算法的改进

论文基于多带谱减法提出了一种改进的单通道语音增强算法研究。对补偿相位谱中的相位补偿函数进行改进,将等效矩形带宽（Equivalent Rectangular Bandwidth）尺度应用于相位补偿函数中,最终把谱减后的语音幅度谱与修正的补偿相位谱相结合得到增强的语音复频谱,而不是保留带噪语音信号的相位谱。对提出的语音增强算法进行性能分析发现,本文提出的算法从主、客观两方面评价均可有效地抑制背景噪声与残余噪声。     

具有高可懂度的改进的维纳滤波的语音增强算法

提出一种具有较高可懂度的基于维纳滤波的语音增强算法。相比于其他语音增强算法,维纳滤波法可以明显提高语音质量且含有较少的音乐噪声,但是它和其他现有语音增强算法一样,都无法有效提高语音可懂度。因为维纳滤波法和其他现有算法都过多注重噪声减少,却忽略了SNR(信噪比)的估计误差和不同的语音幅度谱畸变对可懂度有更重要的影响。为改进这些缺点,此研究依据于先验SNR和增益函数来判定SNR估计误差和语音畸变区域,然后对先验SNR小于-10 d B区域的增益函数进行修正,以及幅度谱畸变大于6.02 d B区域语音进行限制。实验证明,该算法能有效提升增强后语音可懂度NCM(归一化协方差方法)的评测值。     

稀疏低秩模型及相位谱补偿的语音增强算法

针对现有的语音增强算法存在增强效果差、语音信号失真等问题,提出了稀疏低秩模型及改进型相位谱补偿的语音增强算法。首先,用稀疏低秩模型处理含噪语音的幅度谱,得到分离后的语音。接着,用归一化最小均方自适应滤波算法优化相位谱补偿算法的补偿因子。然后,对稀疏低秩分离后的语音进行改进型相位谱补偿处理,得到最终增强的语音。最后,对增强后的语音进行感知语音质量评价分析及频谱分析。实验结果表明,该方法能够有效地去除噪声,并且在低信噪比的情况下,可以保持语音的清晰度。     

Adaptive recurrent nonnegative matrix factorization with phase compensation for Single-Channel speech enhancement

The speech signals are affected by the background noise distortion that is unfavorable to both the intelligibility as well as the speech quality. Most of the speech processing algorithms function with the spectral magnitude without consideration of the spectral phase by leaving them unexplored and unstructured. The proposed single channel speech enhancement model called the Adaptive Recurrent Nonnegative Matrix Factorization (AR-NMF) is designed based on the phase compensation strategy with deep learning. The two major phases considered here are the training phase and the testing phase. During the process of training, the noisy speech signal is decomposed by the Hurst exponent-based Empirical Mode Decomposition (HEMD) and is converted into the frequency domain using Short Time Fourier Transform. Further, the new AR-NMF is used for denoising, where the tuning factor is optimally generated by the optimized RNN. Here, the hidden neurons are optimized using the proposed Adaptive Attack Power-based Sail Fish Optimization (AAP-SFO) with consideration of minimizing the Mean Absolute Error between the actual value and the predicted value. Finally, this phase compensated speech signal is given to the ISTFT that results in the final denoised clean speech signal. From the analysis, the CSED of AAP-SFO-AR-NMF for the street noise is 58.24%, 57.34%, 56.72%, and 77.37% more than RNMF, esHRNR, esTSNR, and Vuvuzela respectively. The performance of the proposed deep enhancement method is extensively evaluated and compared to diverse adverse noisy environments that describe the superiority of the proposed method.

     

Speech enhancement using MMSE estimation under phase uncertainty

Most of the speech enhancement algorithms process the amplitudes of speech, but the phase of noisy speech is left unprocessed as it may cause undesired artifacts. Recently, short time Fourier transform based single channel speech enhancement algorithms are developed by considering uncertain prior knowledge of phase. The uncertain knowledge of the phase is obtained from the phase reconstruction algorithms. The goal of this paper is to develop joint minimum mean square error estimate of complex speech coefficients given uncertainty phase (CUP) information by considering Nagakami probability density function (PDF) and gamma PDF as speech spectral amplitude priors and generalized gamma PDF for noise prior. Estimators like amplitudes given uncertainty phase, which uses uncertain phase only for amplitude estimation and not for phase improvement are developed. Experimental results shows that incorporating uncertain phase information improves quality and intelligibility of speech. Also novel phase-blind estimators are developed using Nagakami PDF/gamma as speech priors and generalized gamma as noise prior. Finally comparison of all estimators using uncertain prior phase information is discussed and how initial phase information affects the enhancement process is analyzed with novel estimators. For comparison of all the derived estimators, the speech signals uttered by male and female speakers are taken from TIMIT database. The proposed CUP estimators outperforms the existing algorithms in terms of objective performance measure segmental signal to noise ratio, phase signal to noise ratio, perceptual evaluation of speech quality, short time objective intelligibility.     

非平稳噪声环境下结合听觉掩蔽的语音增强

针对现有语音增强算法面临残留噪声这一问题,提出一种结合人耳听觉掩蔽的改进算法。将MMSE-LSA谱估计法和一种最优感知增强滤波器融入一个两极语音增强算法框架,利用人耳听觉掩蔽去除残留噪声;给出算法实施的具体步骤和最优感知滤波器的理论推导。实验结果表明,在非平稳噪声环境下,该算法可以有效降低语音失真和残余噪声,提升增强语音信号的主观和客观质量。     

Speech waveform reconstruction from speech parameters for an effective text to speech synthesis system using minimum phase harmonic sinusoidal model for Punjabi

Multimedia Tools and Applications - Speech processing plays a vital role in current speech communication applications. The major objective of digital speech is transmission of messages among human...     

基于APF的谐波补偿中的几种谐波电流检测方法

采用有源电力滤波器（APF）已成为谐波补偿的一种重要趋势，而采用这种方式的关键是能够准确地检测出谐波电流，本文介绍了几种基于APF的谐波补偿中的谐波电流检测方法，并对它们进行了比较。     

相空间重构在语音情感识别中的研究

为了更为全面地表征语音情感状态,弥补线性情感特征参数在刻画不同情感类型上的不足,将相空间重构理论引入语音情感识别中来,通过分析不同情感状态下的混沌特征,提取Kolmogorov熵和关联维作为新的情感特征参数,并结合传统语音特征使用支持向量机（SVM）进行语音情感识别。实验结果表明,通过引入混沌参数,与传统物理特征进行识别的方案相比,准确率有了一定的提高,为语音情感的识别提供了一个新的研究途径。     

A robust adaptive feedforward method for the compensation of harmonic disturbances

Most of disk drives suffer from various disturbances that degrade read–write performances. As the track density rapidly increases, basic drive functions may fail even with a small scale disturbance in a normal operating environment. For the reason, an accurate identification of the repeatable runout (RRO) of a hard disk drive has been one of the most important tasks for the successful servo designs in modern hard disk drive integration. Extensive research efforts have been dedicated for the compensation of the runouts and produced many successful strategies to minimize the influences to the critical basic drive functions. Primarily for the simple implementation and the cost reduction in the actual drive development and manufacturing, most of the developed methods being used in the actual drive integration are preferred to be formulated on time domain or frequency domain with very basic limited functions. The primitive frequency domain approaches usually require extensive calculations and large physical memories. In the present work, a RRO compensation method that combines advantages of the transient Fourier coefficients (TFC) and the least mean square (LMS) update is introduced. Combining the two methods in a proper fashion, the present work provides many benefits for the drive design and outperforms the previous compensation methods. The proposed method requires significantly less amount of computational work and physical memories compared to the conventional runout compensation methods. And it also provides effective frequency component selectivity so that the compensation resources are to be concentrated to a specific problem reason. Comprehensive frequency domain formulation of the method followed by a series of experimental test results is provided in the present article.     

结合语音增强的基音检测改进方法

针对低信噪比情况下,语音信号传统的基音检测方法鲁棒性较差的问题,提出一种结合语音增强的基音检测改进方法。通过基于听觉掩蔽的多频带谱减法减小带噪语音信号背景噪声,得到较干净的语音;将增强后的语音作为基音检测的待处理语音,利用能零积和能零比的多门限法对其进行清浊音判决;在平均幅度差函数（AMDF）加权自相关函数（ACF）的基础方法上进行改进,实现精确的基音检测。理论与仿真结果表明,在信噪比为－10dB时,该方法依然能够精确检测基音周期,鲁棒性明显提高。     

一种基于仿生小波变换的语音增强方法

提出了一种新的基于仿生小波变换的语音增强方法。该方法通过对仿生小波变换系数进行阈值处理,从而达到语音增强的目的。实验结果表明：该方法在四种实际噪声环境下均优于一些经典方法如：谱减法、维纳滤波和基于离散小波变换的阈值去噪方法,具有更好的语音增强效果。