A comparison of subglottal and intraoral pressure measurements during phonation

Intraoral pressure and subglottal pressure, derived from tracheal puncture, were recorded with the electroglottographic signal for one normal speaking male during phonation. The mean subglottal pressure for vowels was also estimated by interpolating the intraoral pressure from surrounding /p/ occlusions. The pressure measurements were highly correlated (r = 0.98) and there were small pressure value differences (on average <2%). The effects of varying speech rate and mode of phonation on the pressure measurements are discussed. A decrease in pressure from the mean subglottal pressure for the open phase and an increase for the closed phase was found during the glottal vibratory cycles.     

IMPLEMENTING THE ADAPTIVE VECTOR QUANTIZER USING CARPENTER/GROSSBERG NET

AVQ(Adaptive Vector Quantizer)overcomes some shortcomings of traditional vectorquantizer with a fixed codebook trained and generated by the LBG or other algorithms by applyinga variab|e codebook.In this paper,we describe an effective and efficient implementation of AVQby modifying the CCN(Carpenter/Grossberg Net).The encoding process of AVQ is very similarto the learning process of the CGN.We study several different encoding schemes,includingwaveform AVQ,analysed parameter AVQ and so on,implemented by the CGN.And we simulatethe encoding performance of each scheme for encoding Gaussian process source,first order Gauss-Markov process source and practical speech signal.Our simulation results show that good qualityboth in subjective and objective tests can be obtained in a low or middle bit rate range.     

基于短时谱最小均方误差估计的语音增强和剩余噪声衰减

本文研究了三种基于语音短时谱最小均方误差估计的语音增强方法:短时谱幅度最小均方误差估计,短时对数谱最小均方误差估计和短时相对谱幅度最小均方误差估计,在理论分析基础上对它们进行了实验研究.计算机仿真结果表明,在加性白色高斯噪声污染下,当带噪语音信噪比为+5～-10dB时,处理后的语音信噪比提高了3.4～12dB。短时对致谱最小均方误差估计的效果最好,试听实验也证实了这一点。 文中还对增强后剩余噪声的衰减问题进行了研究。利用中心削波并对估值器中的增益函数进行修正,可明显地减弱剩余噪声。     

DTW的ASIC实现算法研究

通过分析DTW算法，提出了一种适合ASIC实现的心动阵列结构。仿真结果表明，该并行VLSI处理器阵列系统能够在N M-1个时钟周期内计算出两个模板的匹配加权距离。较之基于通用处理器串行实现的DTW算法需要的3pMN／2个时钟周期，该算法节省了大量的运算时间。     

Is There an Effect of Dysphonic Teachers' Voices on Children's Processing of Spoken Language?

There is a vast body of literature on the causes, prevalence, implications, and issues of vocal dysfunction in teachers. However, the educational effect of teacher vocal impairment is largely unknown. The purpose of this study was to investigate the effect of impaired voice quality on children's processing of spoken language. One hundred and seven children (age range, 9.2 to 10.6, mean 9.8, SD 3.76 months) listened to three video passages, one read in a control voice, one in a mild dysphonic voice, and one in a severe dysphonic voice. After each video passage, children were asked to answer six questions, with multiple-choice answers. The results indicated that children's perceptions of speech across the three voice qualities differed, regardless of gender, IQ, and school attended. Performance in the control voice passages was better than performance in the mild and severe dysphonic voice passages. No difference was found between performance in the mild and severe dysphonic voice passages, highlighting that any form of vocal impairment is detrimental to children's speech processing and is therefore likely to have a negative educational effect. These findings, in light of the high rate of vocal dysfunction in teachers, further support the implementation of specific voice care education for those in the teaching profession.     

Acoustic analysis of voice in multiple sclerosis patients

The objective is to investigate the presence of dysphonic symptoms in multiple sclerosis (MS) patients and to compare quantitative acoustic parameters in multiple sclerosis patients and normal individuals. The method of study was an 8-month controlled cross-sectional that was carried out with 106 individuals (30 MS, 76 controls). Both groups included males and females from 20 to 55 years. Exclusion criteria were prior vocal disorder, laryngeal microsurgery, recent endotracheal intubation, tumors, laryngeal, lung or mediastinal metastases, respiratory disease, and other associated neurological diagnoses. For dysphonic symptoms (qualitative variables), associations were assessed using Mantel-Haenszel's chi2 test, with Yates correction or the Fisher exact test when necessary. Statistical significance was set at p< or =0.05. Dysphonia was observed in 70% of MS individuals versus 33% of controls (p=0.01). Association was found between MS and dysphonia (OR: 2.2, CI 95%: 1.13-4.25). Fundamental frequency was higher among MS patients (p=0.01). Fundamental frequency deviation was significantly higher in MS women (but not men) than controls (p=0.00). Jitter was higher in MS men than in all other groups (p=0.00). Results suggest that evaluation and treatment of MS patients should be revised, evaluating voice alterations in relation to other signs. MS seems to intensify gender effect on fundamental frequency deviation, noise, and jitter, with MS women presenting fewer voice variations than men.     

语音信号互信息估计的非线性搜索算法及识别应用

基于互信息理论的语音识别方法不仅考虑了语音信号的时变分布特征，并且考虑了语音信号的统计分布特征，能有效地提高同类模式的凝聚度，减少非同类模式间的耦合性，在语音识别实验和实际应用中反映出良好的识别精度和很高的运行效率，与其它方法相比更适合嵌入式系统的语音识别应用。本文提出了一种互信息估计的非线性搜索算法，这一算法能够有效地处理语音信号时变分布特征的非线性波动，进一步提高语音模式互信息匹配的精度。     

电脑语音系统的集成与设计

电脑语音集成技术结合了通信技术和计算机技术，近年来发展迅速，本文指出ＣＴＩ系统的集成与设计应考虑ＣＴＩ标准，语音卡，硬件平台，操作系统，语音处理，系统性能等方面的问题。     

在帧中继骨干网上传送话音的技术

在帧中继网上传送话音（ＶｏＦＲ）是帧中继网上的一项增值业务。本文提出该业务的主要障碍是话音分组的时延和丢弃；接着又介绍了克服这些障碍的技术，包括话音压缩、采用回波消除器和抖动缓冲器以及优先级技术等；最后又探讨了ＶｏＦＲ的兼容性和标准化问题。     

语音识别中的神经网络DTW结构

本文介绍语音识别中一种基于动态规划技术的时间规正算法DTW的神经网络实现方法.DTW是语音识别中最为有效的方法之一,它具有较强的鲁棒性且为语音识别系统提供了可能的最高识别率.但由于其计算量太大,除非用专门的硬件,DTW算法在实现时受到了限制.在本文中,所有的计算是由两个循环神经子网和一记忆层来完成的,该方法展示了算法的硬接线结构,(hard-wiring)的优越性,这为DTW的硬件实现提供了一种新的实施策略.