人脸语音动画中基于PSOLA的情感语音合成系统

提出一种基于时域基音同步叠加TD-PSOLA算法的情感语音合成系统。根据情感语音库分析总结情感规则,在此基础上利用TD-PSOLA算法对中性语音的韵律参数进行改变,并提出一种能够对基频曲线尾部形状改变的方法,使句子表达出丰富的情感。实验表明,合成出的语音具有明显的情感色彩,证明了该系统能以简单明了的方式实现情感语音的合成,有助于提高人脸语音动画表达的丰富性和生动性。     

Prosody conversion from neutral speech to emotional speech

Emotion is an important element in expressive speech synthesis. Unlike traditional discrete emotion simulations, this paper attempts to synthesize emotional speech by using "strong", "medium", and "weak" classifications. This paper tests different models, a linear modification model (LMM), a Gaussian mixture model (GMM), and a classification and regression tree (CART) model. The linear modification model makes direct modification of sentence F0 contours and syllabic durations from acoustic distributions of emotional speech, such as, F0 topline, F0 baseline, durations, and intensities. Further analysis shows that emotional speech is also related to stress and linguistic information. Unlike the linear modification method, the GMM and CART models try to map the subtle prosody distributions between neutral and emotional speech. While the GMM just uses the features, the CART model integrates linguistic features into the mapping. A pitch target model which is optimized to describe Mandarin F0 contours is also introduced. For all conversion methods, a deviation of perceived expressiveness (DPE) measure is created to evaluate the expressiveness of the output speech. The results show that the LMM gives the worst results among the three methods. The GMM method is more suitable for a small training set, while the CART method gives the better emotional speech output if trained with a large context-balanced corpus. The methods discussed in this paper indicate ways to generate emotional speech in speech synthesis. The objective and subjective evaluation processes are also analyzed. These results support the use of a neutral semantic content text in databases for emotional speech synthesis.     

Speech emotion recognition: Features and classification models

To solve the speaker independent emotion recognition problem, a three-level speech emotion recognition model is proposed to classify six speech emotions, including sadness, anger, surprise, fear, happiness and disgust from coarse to fine. For each level, appropriate features are selected from 288 candidates by using Fisher rate which is also regarded as input parameter for Support Vector Machine (SVM). In order to evaluate the proposed system, principal component analysis (PCA) for dimension reduction and artificial neural network (ANN) for classification are adopted to design four comparative experiments, including Fisher + SVM, PCA + SVM, Fisher + ANN, PCA + ANN. The experimental results proved that Fisher is better than PCA for dimension reduction, and SVM is more expansible than ANN for speaker independent speech emotion recognition. The average recognition rates for each level are 86.5%, 68.5% and 50.2% respectively.     

A decision support system based on fuzzy reasoning and AHP–FPP for the ecodesign of products: Application to footwear as case study

The design stage represents one of the most critic steps for product development. Here, a great number of considerations have to be borne in mind, e.g., technical, functional, aesthetic or economic criteria. More recently, the increasing concerns on environmental aspects have added complexity to the process, known as ecodesign. In this respect, a framework to integrate the criteria provided by quantitative environmental indicators has been proposed on the basis of Fuzzy Preference Programming method features and fuzzy logic reasoning. As a result, an integrated Ecodesign Index (EcoInd) is obtained. This idea enables the decision making at process and product level taking into account different indicators at a time. The ecodesign of children's footwear was taken as case study and an ecodesign tool (decision support system) that included the estimation of environmental indicators and their integration was developed. Different models of shoes were analyzed to identify the most environmentally friendly design and to test the tool. In this case, the Ecological Footprint and two Environmental Risk Assessment indicators, namely Hazard Quotient and Cancer Risk, were selected as relevant environmental indicators and they were computed from data provided by a shoes manufacturer. Then, these indicators were integrated in the ecodesign tool and the EcoInd values were appraised for the children's footwear models analyzed. According to these figures, they were ranked as Red Leather > White Leather > White Synthetic > Pink Synthetic, from best to worst.     

Modeling perspective projections in 3-dimensions by rotations in 4-dimensions

We show how to represent perspective projections in 3-dimensions using rotations in 4-dimensions. This representation permits us to replace classical singular 4 × 4 matrices for perspective projection with nonsingular 4 × 4 orthogonal matrices. This approach also allows us to compute perspective projections by sandwiching vectors between two copies of a unit quaternion. In addition to deriving explicit formulas for these 4 × 4 rotation matrices for perspective projection, we also explain the geometric intuition underlying the observation that perspective projections in 3-dimensions can be represented by rotations in 4-dimensions. We show too that every rotation in 4-dimensions models either a rotation, a reflection, a perspective projection, or one of their composites in 3-dimensions.     

A prosodic phrasing model for a Korean text-to-speech synthesis system

This paper presents a prosodic phrasing model for Korean to be used in a text-to-speech synthesis (TTS) system. Read text corpora were morpho-syntactically parsed and prosodically labeled following the Penn Korean Treebank (Han, Chunghye, Ko, Eon-Suk, Yi, Heejong, Palmer, M., 2002. Penn Korean Treebank: development and evaluation. In: Proceedings of the 16th Pacific Asian Conference on Language and Computation. Korean Society for Language and Information.) and K-ToBI prosodic labeling conventions (Sun-Ah, J., 2000. K-ToBI (Korean ToBI) labelling conventions. Version 3.1. Available from: URL <http://www.linguistics.ucla.edu/people/jun/ktobi/K-tobi.html>.), respectively. Decision trees were trained with morpho-syntactic and textual distance features to predict locations of accentual and intonational phrase breaks. Our phrasing model cross-validated on a 300-sentence corpus (6936 words or 21,436 syllables, with an average of 72 syllables or 23 words per sentence) predicted non-breaks with F = 92.4% and breaks with F = 88.0% (F = 72.8% for accentual phrase breaks and F = 71.3% for intonational phrase breaks).     

基于OCC模型和贝叶斯网络的情绪句分类方法

情绪句分类是情绪分析研究领域的核心问题之一,旨在解决情绪句类别的自动判断问题。传统基于情绪认知模型(OCC模型)的情绪句分类方法大多依赖词典和规则,在文本信息缺失的情况下分类精度不高。文中提出基于OCC模型和贝叶斯网络的情绪句分类方法,通过分析OCC模型的情绪生成规则,提取情绪评估变量并结合情绪句中含有的表情符号特征构建情绪分类贝叶斯网络;通过概率推理,可以实现句子级文本的情绪分类,并减小句中信息缺失所带来的影响。与NLPCC2014中文微博情绪分析评测的子任务情绪句分类评测结果的对比表明,所提方法具有有效性。     

Improved fault detection employing hybrid memetic fuzzy modeling and adaptive filters

We propose an improved fault detection (FD) scheme based on residual signals extracted on-line from system models identified from high-dimensional measurement data recorded in multi-sensor networks. The system models are designed for an all-coverage approach and comprise linear and non-linear approximation functions representing the interrelations and dependencies among the measurement variables. The residuals obtained by comparing observed versus predicted values (i.e., the predictions achieved by the system models) are normalized subject to the uncertainty of the models and are supervised by an incrementally adaptive statistical tolerance band. Upon violation of this tolerance band, a fault alarm is triggered. The improved FD methods comes with two the main novelty aspects: (1) the development of an enhanced optimization scheme for fuzzy systems training which builds upon the SparseFIS (Sparse Fuzzy Inference Systems) approach and enhances it by embedding genetic operators for escaping local minima ⟶ a hybrid memetic (sparse) fuzzy modeling approach, termed as GenSparseFIS. (2) The design and application of adaptive filters on the residual signals, over time, in a sliding-window based incremental/decremental manner to smoothen the signals and to reduce the false positive rates. This gives us the freedom to tighten the tolerance band and thus to increase fault detection rates by holding the same level of false positives. In the results section, we verify that this increase is statistically significant in the case of adaptive filters when applying the proposed concepts onto four real-world scenarios (three different ones from rolling mills, one from engine test benches). The hybridization of sparse fuzzy inference systems with genetic algorithms led to the generation of more high quality models that can in turn be used in the FD process as residual generators. The new hybrid sparse memetic modeling approach also achieved fuzzy systems leading to higher fault detection rates for some scenarios.     

基于基频特征的情感语音识别研究*

探索在不同的情感状态下的基音特征变化规律.通过对含有生气、高兴、悲伤情感语音信号进行分析,总结了情感语音基频的变化规律,确定了用于情感识别的12维的基频的基本特征以及扩展特征,运用混合高斯模型进行情感识别,并作了识别实验,获得了较好的结果.     

Optimal selection of full scale tannery effluent treatment alternative using integrated AHP and GRA approach

The selection of optimal wastewater treatment alternative involves multiple objectives and/or criteria and hierarchy process. This study integrates analytical hierarchy process (AHP) and grey relation analysis (GRA) for optimal selection of full scale tannery effluent treatment plants. For this purpose, seven tanneries and their effluent treatment facilities are studied in detail in Southern India. The objective hierarchy criterion is considered based on three factors; economic, technical, and administrative, each of which again involves hierarchy of indices. A realistic treatment alternative selection is obtained since all the data used is on actual basis. The biggest advantage of this approach is that it provides the information regarding the scope for further improvement in existing treatment options. The study indicates that the AHP and grey relation analysis are powerful tools that can be used for implementation of appropriate wastewater treatment technology.     

Characterization of a capacitive tactile shear sensor for application in robotic and upper limb prostheses

This paper presents the characterization of a novel tactile sensor designed to measure shear forces. The sensor design is targeted for use in robotic and prosthetic hands, where haptic feedback or ability to detect shear forces associated with slip are critical. The presented sensor utilizes the principle of differential capacitance to measure the mechanical deflection of the sensor element. The dynamic range of the sensor can be varied by encapsulating the sensor terminal within silicone of varying hardness. The design features ease of mass production, low per-unit-cost, novel overload protection and low wire count, while still preserving the ability to achieve reasonable spatial resolutions and array densities. Mathematical and COMSOL multiphysics models of the sensor are presented, in addition to results from practical experiments. Sensors with a full scale displacement range of ±0.525 mm were produced and the differential capacitance was measured. Shear force transduction was characterized over the range of 0 N–4 N with the sense terminal encapsulated by silicone with a shore A hardness of 20. The effect of elastomer hardness on the sensor's dynamic range was analyzed. The differential capacitance, when measured at each fixed interval, was found experimentally to have a maximum standard deviation of 4.28e?16 F over a ±2 N range. A maximum standard deviation of 1.35e?15 F was measured across characterized full scale sensor range of ±4 N. The sensor design has a sensitivity of 1.967 fF/N of applied force and the sensor output was found to be approximately linear. The coefficient of determination, r², was found to be 0.941.     

Fuzzy model tuning using simulated annealing

This paper presents the use of simulated annealing metaheuristic for tuning Mamdani type fuzzy models. Structure of the Mamdani fuzzy model is learned from input–output data pairs using Wang and Mendel’s method and fuzzy c-means clustering algorithm. Then, parameters of the fuzzy system are tuned through simulated annealing. In this paper, we perform experiments to examine effects of (a) initial solution generated by Wang and Mendel’s method and fuzzy c-means clustering method, (b) membership function update procedure, (c) probability parameter for the calculation of the initial temperature, (d) temperature update coefficient used for cooling schedule, and (e) randomness level in the disturbance mechanism used in simulated annealing algorithm on the tuning of Mamdani type fuzzy models. Experiments are performed with Mackey–Glass chaotic time series. The results indicate that Wang and Mendel’s method provides better starting configuration for simulated annealing compared to fuzzy c-means clustering method, and for the membership function update parameter, MFChangeRate ∈  (0, 1], and the probability parameter for the calculation of the initial temperature, P₀ ∈  (0, 1), values close to zero produced better results.     

Short-term non-uniform access in IEEE 802.11-compliant WLANs: A study on its impact on the saturation performance

Along with the success of IEEE 802.11-compliant WLANs, the distributed coordinated function (DCF) specified as the contention-based medium access mechanism in IEEE 802.11 has been widely used to support applications in new regimes. A thorough understanding of the characteristics of IEEE 802.11 DCF is therefore a research focus. DCF assumes the binary exponential backoff algorithm (BEBA) [D. Bertsekas, R. Gallager, Data Networks, Prentice Hall, 1992]. Several existing models that characterize BEBA as a p-persistent scheme have ignored several subtle protocol details. This has practically constrained the models from being applied in a broader range of configurations, e.g., small contention window, and from being further extended, e.g., 802.11e EDCA.In this paper, we develop an analytical model that captures the subtlety, and faithfully describes the channel activities as governed by DCF. Based on the devised model, we perform a rigorous analysis on the saturation throughput performance in a single-hop WLAN. We show that the stochastic property of the backoff time, r, may substantially impact the system performance. For example, setting the range of r to [1, CW] instead of [0, CW − 1], can degrade the system throughput considerably. We also identify, by clearly defining and thus being able to differentiate the two terms, attempt probability and transmission probability, an erroneous extension made to Bianchi’s model [G. Bianchi, Performance analysis of the IEEE 802.11 distributed coordination function, IEEE JSAC, 18(3) (2000) 535–547]. All the findings are corroborated by ns-2 simulation.     

Spider specie identification and verification based on pattern recognition of it cobweb

Biodiversity conservation is a global priority where the study of every type of living form is a fundamental task. Inside the huge number of the planet species, spiders play an important role in almost every habitat. This paper presents a comprehensive study on the reliability of the most used features extractors to face the problem of spider specie recognition by using their cobwebs, both in identification and verification modes. We have applied a preprocessing to the cobwebs images in order to obtain only the valid information and compute the optimal size to reach the highest performance. We have used the principal component analysis (PCA), independent component analysis (ICA), Discrete Cosine Transform (DCT), Wavelet Transform (DWT) and discriminative common vectors as features extractors, and proposed the fusion of several of them to improve the system’s performance. Finally, we have used the Least Square Vector Support Machine with radial basis function as a classifier. We have implemented K-Fold and Hold-Out cross-validation techniques in order to obtain reliable results. PCA provided the best performance, reaching a 99.65% ± 0.21 of success rate in identification mode and 99.98% ± 0.04 of the area under de Reveicer Operating Characteristic (ROC) curve in verification mode. The best combination of features extractors was PCA, DCT, DWT and ICA, which achieved a 99.96% ± 0.16 of success rate in identification mode and perfect verification.     

The social sharing of emotion (SSE) in online social networks: A case study in Live Journal

Social Sharing of Emotion (SSE) occurs when one person shares an emotional experience with another and is considered potentially beneficial. Though social sharing has been shown prevalent in interpersonal communication, research on its occurrence and communication structure in online social networks is lacking. Based on a content analysis of blog posts (n = 540) in a blog social network site (Live Journal), we assess the occurrence of social sharing in blog posts, characterize different types of online SSE, and present a theoretical model of online SSE. A large proportion of initiation expressions were found to conform to full SSE, with negative emotion posts outnumbering bivalent and positive posts. Full emotional SSE posts were found to prevail, compared to partial feelings or situation posts. Furthermore, affective feedback predominated to cognitive and provided emotional support, empathy and admiration. The study found evidence that the process of social sharing occurs in Live Journal, replicating some features of face to face SSE. Instead of a superficial view of online social sharing, our results support a prosocial and beneficial character to online SSE.     

Combining AVHRR and meteorological data for estimating live fuel moisture content

Spatial assessment of fire risk is very important for reducing the impacts of wildland fires. Several variables related to fire ignition, propagation and its effects are included in fire risk analysis. Life Fuel Moisture Content (LFMC) is one such variable, which is highly related to fire ignition, and propagation. A wide variety of methods have been applied to estimate LFMC, including field sampling and meteorological indices. Given the limitations of these methods, satellite images are a sound alternative for estimating LFMC because of their capability to spatially and temporally monitor the vegetation status.This paper aims to improve previous empirical models to estimate LFMC from satellite images, by considering meteorological information. The original models proposed by Chuvieco et al. [Chuvieco, E., Cocero, D., Riaño, D., Martin, M.P., Martinez-Vega, J., et al., (2004). Combining NDVI and surface temperature for the estimation of live fuel moisture content in forest fire danger rating. Remote Sensing of Environment, 92, 322–331] for grasslands and shrublands were used as starting point for this research. High over-estimation of LFMC values from those models was found when applied to dry years. Consequently, the new models proposed in this paper use a simple drought index to discriminate between dry and wet years at the beginning of the spring season. A different harmonic function was fitted to each group of hydrological years, to take into account the inter-annual variations in LFMC seasonal trends. Subsequently, two empirical models, one for grasslands and one for shrubs (Cistus ladanifer), were derived based on multivariate linear regression analysis of the data collected at Cabañeros National Park (Central Spain). Determination coefficients greater than 0.8 for grasslands and 0.7 for shrubs were found. The models showed good performance too when applied to other plots of grasslands (R² = 0.76) and shrubland (R² = 0.71) with similar environmental characteristics to the calibration site.     

Appearance models based on kernel canonical correlation analysis

This paper introduces a new approach to constructing appearance models based on kernel canonical correlation analysis (kernel-CCA). Kernel-CCA is a non-linear extension of CCA, where a non-linear transformation of the input data is performed implicitly using kernel methods. Although, in this respect, it is similar to other generalized linear methods, kernel-CCA is especially well suited for relating two sets of measurements. The benefits of our method compared to standard feature extraction methods based on PCA will be illustrated experimentally for the task of estimating an object's pose from raw brightness images.     

Choosing among weight-estimation methods for multi-criterion analysis: A case study for the design of multi-purpose offshore platforms

Application of the sustainability concept to environmental projects implies that at least three feature categories (i.e., economic, social, and environmental) must be taken into account by applying a participative multi-criterion analysis (MCA). However, MCA results depend crucially on the methodology applied to estimate the relative criterion weights. By using a logically consistent set of data and methods (i.e., linear regression [LR], factor analysis [FA], the revised Simos procedure [RSP], and the analytical hierarchy process [AHP]), the present study revealed that mistakes from using one weight-estimation method rather than an alternative are non-significant in terms of satisfaction of specified acceptable standards (i.e., a risk of up to 1% of erroneously rejecting an option), but significant for comparisons between options (i.e., a risk of up to 11% of choosing a worse option by rejecting a better option). In particular, the risks of these mistakes are larger if both differences in statistical or computational algorithms and in data sets are involved (e.g., LR vs. AHP). In addition, the present study revealed that the choice of weight-estimation methods should depend on the estimated and normalised score differences for the economic, social, and environmental features. However, on average, some pairs of weight-estimation methods are more similar (e.g., AHP vs. RSP and LR vs. AHP are the most and the least similar, respectively), and some single weight-estimation methods are more reliable (i.e., FA > RSP > AHP > LR).