录音教材产品录制加工中的语音信号响度平衡

【】响度平衡是语言录音教材产品录制加工中的重要工作,但也是常被忽视的环节。本文通过对人耳主观响度的特征、主观响度的计量与监控以及录音环节中影响信号响度的因素等多个角度的分析,深入探讨了语言录音教材产品录制加工中响度平衡的原则和方法,力图为高质量的语音教材的生产提供有力的参考。     

一种提高嵌入率的语音融合隐藏改进算法

语音信号在较高的采样率下,采样数据具有冗余度。对于秘密语音信号,冗余度越大,算法的嵌入率越低。本文结合信号抽取插值方法,对已有的语音融合隐藏算法提出进一步的改进,以减小秘密语音的冗余度,提高算法的嵌入率。利用高质量的计算机网络这一信息传输途径,进行实验仿真,并从主客观两方面对算法的性能进行评估。实验结果表明,利用本文改进算法,秘密语音的嵌入率得到提高,且恢复的秘密语音信号质量较好,PESQ打分在4.0左右,可懂度高。算法具有较高的可行性。     

基于改进型双门限语音端点检测算法的研究

语音端点检测是语音信号处理过程中的一个重要步骤,其检测准确性直接影响语音信号处理的速度和效果.传统的基于双门限法语言检测技术,在语音处于纯语音情况下判断语音端点较准确,但在语音处于噪声情况下,尤其是低信噪比的情况下,端点识别率很低,出错率很高.基于提高此方法识别率的目的,采用调整阈值个数,平滑滤波,引入语音结束最小长度的方法对其进行改进,通过了Matlab仿真实验,得出了较好的语音端点检测准确率.     

基于循环神经网络的汉语语言模型建模方法

语言模型是语音识别系统的重要组成部分,目前的主流是n-gram模型。然而n-gram模型存在一些不足,对语句中长距信息描述差、数据稀疏是影响模型性能的两个重要因素。针对不足,研究者提出循环神经网络(Recurrent Neural Network,RNN)建模技术,在英语语言模型建模上取得了较好的效果。根据汉语特点将RNN建模方法应用于汉语语言建模,并结合两种模型的优点,提出了模型融合构建方法。实验结果表明:相比传统的n-gram语言模型,采用RNN训练的汉语语言模型困惑度(Per PLexity,PPL)有了下降,在对汉语电话信道的语音识别上,系统错误率也有下降,将两种语言模型融合后,系统识别错误率更低。     

Ultrastructure of dermal denticles in sharpnose shark (Rhizoprionodon lalandii) (Elasmobranchii,Carcharhinidae)

The presence of denticles in the external surface, oral cavity, fins, and clasper of Elasmobranchii has been widely reported. These structures, called body denticles, may be observed on the body surface of sharks. Dermal and oral denticles are made up by a basal plate that is embedded in the dermis, forming a peduncle that grows from the base to the crown. These denticles may protect the skin against abrasion, and improve hydrodynamics and gill arches function. Rhizoprionodon lalandii is a widely distributed and very common species in Brazilian coastal areas. The aims of this study was to compare the morphology of oral and body denticles of R. lalandii to understand the implications of these structures in the behavior of these animals. Morphological analysis showed that there are differences between dermal and oral denticles, which are related to their role in different body regions. Body denticles have three cusps, and well‐defined crests and ridges, and literature data suggest that suggest that hydrodynamics is their main function. Most of the oral denticles have only one cusp, and their morphology and distribution showed that their main functions are preparing food to be swallowed and protecting the oral cavity against abrasion. Microsc. Res. Tech. 78:859–864, 2015. © 2015 Wiley Periodicals, Inc.     

Research on nonlinear dynamics features of emotional speech

The application of nonlinear measures based on the chaotic characteristics of emotional speech is proposed. Nonlinear features such as minimum delay time, dimension correlation, Kolmogorov entropy, Lyapunov exponent and Hurst exponent are extracted from the emotional speech signal. The performance of nonlinear features is verified by the comparisons of recognition rates of different features (nonlinear characteristics, prosodic features and MFCC features). First, the Berlin emotional speech database and TYUT2.0 emotional speech database are chosen as the corpus independently, both covering three emotional classifications (anger, happiness and fear). The effectiveness of the nonlinear characteristics is tested on the Support Vector Machine Network. The result shows that the performance of nonlinear features outperforms that of prosodic features on the Berlin emotional speech database and that of prosodic features and MFCC on TYUT2.0 emotional speech database. In addition, nonlinear features have obvious advantage in detecting more natural emotional speech and better robustness.     

腭裂修复术对婴幼儿听力和中耳功能的短期影响

目的　探讨单纯腭裂修复术对腭裂婴幼儿听力和中耳功能的短期影响。方法　2013年1月至2014年11月在北京协和医院行手术治疗的先天性腭裂婴幼儿33例，其中男22例，女11例，在腭裂修复术前、术后1~24个月对患儿进行畸变产物耳声发射（distortion product otoacoustic emission, DPOAE）、听性脑干反应（auditory brainstem response，ABR）和226 Hz鼓室声导抗图检测，观察腭裂修复术后患儿听力和中耳功能的短期变化。结果　33例（66耳）腭裂婴幼儿手术时平均年龄为17个月（8.5 ~25个月）。64耳接受术前DPOAE测试，7耳（10.9%）通过；56耳接受术后DPOAE测试，19耳（33.9%）通过，手术前后通过率比较差异具有统计学意义（P＜0.01）；术前术后均接受DPOAE测试的54耳中，16耳（29.6%）听力改善。63耳接受术前ABR检查，其中Ⅴ波反应阈平均值为（43.73±13.65）dB nHL；62耳接受术后ABR检查，其中Ⅴ波反应阈平均值为（35.65±13.75）dB nHL，手术前后反应阈值比较差异具有统计学意义（P＜0.01）；术前术后均接受ABR检测的53耳中，29耳（54.7%）术后V波反应阈降低，听力改善。手术前后226 Hz鼓室声导抗图分型比较差异无统计学意义（P＞0.05）。结论　单纯腭裂修复术术后患儿短期听力得到改善，但中耳功能改善不明显。     

Lattice Vector Quantization Applied to Speech and Audio Coding

Lattice vector quantization(LVQ) has been used for real-time speech and audio coding systems.Compared with conventional vector quantization,LVQ has two main advantages:It has a simple and fast encoding process,and it significantly reduces the amount of memory required.Therefore,LVQ is suitable for use in low-complexity speech and audio coding.In this paper,we describe the basic concepts of LVQ and its advantages over conventional vector quantization.We also describe some LVQ techniques that have been used in speech and audio coding standards of international standards developing organizations(SDOs).     

Towards building a social emotion detection system for online news

Social emotion detection of online users has become an important task for mining public opinions. Social emotion detection aims at predicting the readers’ emotions evoked by news articles, tweets, etc. In this article, we focus on building a social emotion detection system for online news. The system is built based on the modules of document selection, Part-of-speech (POS) tagging, and social emotion lexicon generation. Empirical studies are extensively conducted on a large scale real-world collection of news articles. Experiments show that the document selection algorithm has a positive effect on the social emotion detection. The system performs better with the words and POS combination compared to a feature set consisting only of words. POS is also useful to detect emotion ambiguity of words and the context dependence of their sentiment orientations. Furthermore, the proposed method of generating the lexicon outperforms the baselines in terms of social emotion prediction.     

Speech and noise power estimation using Gamma modeling

In speech enhancement, having an accurate estimation of the power of the speech and noise signals forming the noisy observation is critical, as it can highly affect the performance of the enhancement algorithm. A method is introduced in which the distributions of the power of the speech and noise periodograms are modeled using the Gamma distribution to extract their shape parameters. These shape parameters are later used in the observed noisy speech to estimate the power when forming speech and noise periodograms. This method results in more accurate and faster power estimation with respect to the well‐known minimum statistics power estimation algorithm and together with the maximum a posteriori speech enhancement algorithm exhibits good speech enhancement performance.