Search

Breast cancer is a neoplastic disease that can be diagnosed either as benign or malign according to the growth-rate of the neoplastic lesion. Owing to the relevance of obtaining better detection tools, this work describes the development and optimization of support vector machines for the classification of the types of such cancer. Tests were performed using the breast cancer dataset of the University of Wisconsin Hospitals, USA, available at the Machine Learning Repository of the University of California Irvine. The radial basis function kernel was selected for the classifier and its hyperparameters were refined using two methods: particle swarm optimization and genetic algorithms. The results for the first method exhibited 97.71% accuracy, 96.30% sensitivity, and 98.65% of selectivity. On the other hand, using the second method, the accuracy was 95.78%, with sensitivity and selectivity of 96.73% and 95.25%, respectively. Therefore, there is an indication that these search algorithms are viable tools to optimize machine learning models for the purpose of breast cancer classification.

Induction motors play a major role in the industry nowadays due to their simple construction, uncomplicated maintenance, and cost efficiency. As the motor operates for repeated hours, some faults may occur and, depending on the process sensitivity, can cause significant losses to the industrial production. In this context, an alternative method is proposed to detect and classify bearing faults using acoustic emission signals generated by the machines and simple features obtained from them. A pair of condenser microphones was used to acquire these signals and audio feature extraction is performed to obtain time and frequency patterns to characterize healthy and faulty machines. The major differences of the acoustic signals regarding the fault frequency signatures are discussed by analyzing specific peaks observed in their spectra. Several coupled load values and levels of power supply voltage unbalance were considered in the experimental tests, which emulate common situations encountered in industrial environments, obtaining accuracy rates over 97% of success. Finally, a comparison is presented of machine learning techniques for bearing faults classification under different load values and voltage unbalance levels.

Characteristics like robustness, adaptability to several load conditions, and low costs of operation and maintenance are some of the reasons three-phase induction motors are ubiquitous in industrial applications. Even so, these machines are subjected to electrical and mechanical faults which can result in malfunction. The detection of incipient faults is the focus of several recent studies because they provide information for timely decisions to avoid unplanned stops of industrial processes. A common problem of induction motors is the presence of broken rotor bars. Most of the methods of condition monitoring and fault diagnosis focus on the usage of only one type of power supply: line-connected or fed by frequency inverters. Given this situation, we present an alternative technique for the detection of broken rotor bars of three-phase induction motors regardless of the type of power supply. Our approach is based on similarity measures of the stator current signals of two motor phases in order to extract the relevant features of such signals, which are then classified using three intelligent systems: artificial neural network, support vector machines, and k-nearest neighbors. We performed 3310 experimental tests with motors operating in steady state, with sinusoidal and non-sinusoidal power supply, and variations of voltage unbalance levels, supply frequency, and load torque. Classification accuracy rates over 92% of success were obtained in these tests, which validate the proposed approach.

This paper proposes a novel approach for detection of stator short-circuit faults in three-phase induction motors. The method is based on two stages: feature extraction and classification by intelligent systems. First, mutual information is estimated from delayed stator current signals, which are used as inputs of C4.5 decision trees and multilayer perceptron neural networks in the second step. Several offline and online experimental tests are presented considering voltage unbalance, load torque variations, and 1% to 10% short-circuit levels. The obtained results corroborate the effectiveness of this new diagnostic approach.

A análise acústica é uma técnica de processamento digital de sinais, não invasiva, capaz de auxiliar na avaliação de distúrbios da voz. Um dos métodos utilizados nesta análise acústica consiste em obter características que discriminem os sinais de forma automática. Portanto apresenta-se uma proposta que utiliza como sinal de entrada a vogal "a" sustentada para extrair informações relevantes dos sinais. A extração de informação é obtida pela transformada wavelet, seguida do cálculo de energia e entropia dos coeficientes resultantes. Por fim estas características são classificadas pelo classificador support vector machine. A classificação das classes saudáveis e patológicos obteve um acerto médio de 94.32% para a grandeza energia, 93.83% para a grandeza de entropia e com as grandezas juntas 94.32%. Para a classificação apenas das patologias obteve um acerto médio de 96.30% para a grandeza energia, 95.93% para a grandeza entropia e 96.67% para as grandezas juntas.

The use of body signals for health care applications has become ubiquitous in the last decade. One utilization of such measurements is the monitoring of respiratory flow for physiotherapy assistance. This application is based on relative flow measures which can rely on inexpensive sensors. Based on that, we present a low-cost electronic device that detects blows and suctions with a pressure sensor and emulates a keyboard for interfacing with computers. This joystick allows children to control free internet games by associating blows and suctions with different intensities to keyboard actions. Also, the intensity can be calibrated according to the user's pulmonary capacities. This feature is adequate for gradual respiratory physiotherapy and can be customized for each patient. In order to verify the operation of the proposed device, practical tests were performed with three online free games, where the joystick functionality was assessed with different therapeutic configurations.

This work analyses the performance-complexity tradeoff for different direction of arrival (DoA) estimation techniques. Such tradeoff is investigated taking into account uniform linear array structures. Several DoA estimation techniques have been compared, namely the conventional Delay-and-Sum (DS), Minimum Variance Distortionless Response (MVDR), Multiple Signal Classifier (MUSIC) subspace, Estimation of Signal Parameters via Rotational Invariance Technique (ESPRIT), Unitary-ESPRIT and Fourier Transform method (FT-DoA). The analytical formulation of each estimation technique as well the comparative numerical results are discussed focused on the estimation accuracy versus complexity tradeoff. The present study reveals the behavior of seven techniques, demonstrating promising ones for current and future location applications involving DoA estimation, especially for 5G massive MIMO systems., 35 pages, 10 figures, 4 tables

In this paper, a power allocation scheme based on an evolutionary heuristic approach, namely Firefly Algorithm (FA) is proposed for the mitigation of fiber temperature fluctuations effects in the optical code division multiplexing access (OCDMA) networks. The temperature fluctuations degrade the 2-D wavelength-hopping time spreading optical codes by inducing a distortion on the autocorrelation of the received signal. The fiber temperature fluctuations are ordinarily hard to accurately determine and compensate due to the dynamic nature of the environmental temperature variation. In this context, power allocation (PA) policies constitute an efficient way to dynamically mitigate the effects of temperature variation with low cost and complexity of implementation. In this work, an FA input parameter optimization procedure has been conducted aiming to guarantee an efficient FA-based OCDMA power allocation algorithm regarding convergence velocity and quality of the solutions trade-off. The numerical results have demonstrated the effectiveness of the proposed FA power allocation scheme in mitigating the effects of fiber temperature fluctuations, as well as balancing the near–far effect (NFE) and multiple access interference (MAI). Moreover, the influence of the code parameters and the number of the nodes on the FA-based OCDMA power allocation scheme has been investigated. The comparison from both evolutionary heuristics FA and particle swarm optimization (PSO) power allocation schemes has indicated the faster convergence of FA-based scheme when the number of nodes K increases, while for both the complexity is similar, resulting in polynomial order O ( K 2 ) . Moreover, the FA-based power allocation approach presents a smoother and monotonic convergence when compared with the more oscillatory behavior of the PSO-based power allocation procedure.

The three-phase induction motors are considered one of the most important elements of the industrial process. However, in this environment, these machines are subject to electrical and mechanical faults, which may cause significant financial losses. Thus, the purpose of this paper is to present a pattern recognition method for the detection of stator windings short circuits based on measures of mutual information between the phase current signals. In order to validate the proposed patterns, feature vectors obtained from normal and faulty motors are applied to two topologies of artificial neural networks. The classification results presented accuracies over 93% even when the motors were subject to several conditions of load torque and power supply voltage unbalance.

This paper proposes the minimization of the distortion measure of signals reconstructed by wavelets using genetic algorithm. The minimization happens between the original study object and the reconstructed one—Percent Root-Mean-Square. This distortion measure is analyzed in wavelet creation that are responsible of signals reconstruction with the best approximation from its originals, and these signals, used as study object, are electrocardiograms signals. This proposal solves the problem main commonly found in signals compressing area which is exactly the reconstruction of signals without loss of informations. The proposed method provided satisfactory results compared to those found in the literature.

This paper proposes a method for classification of broken rotor bars of three-phase induction motors driven with three different models of frequency inverters by using the amplitude of the stator current signal in the time domain. To assess the effectiveness of the method under evaluation in this work, classification accuracy of four different learning machine techniques were compared: Fuzzy ARTMAP network; Support Vector Machine; k-Nearest Neighbour and Multilayer Perceptron Network. Moreover, this paper presents the comparison of an analogue and a digital filter to the line current signal provided by the motor. Results obtained from 2293 experimental tests are presented to validate this comparative study, considering a wide range of speed and load conditions.

Maquinas eletricas sao alvos de pesquisa sobre varias tecnicas quando envolve a deteccao e prognostico de falhas. Apesar de amplamente consolidados, muitos metodos dependem de sensores que necessitam parada no funcionamento do motor para que possam ser instalados, o que acarreta em prejuizos em cenarios industriais. Outra opcao e a instalacao destes sensores assim que a maquina e dimensionada e instalada, o que exige planejamento. Neste trabalho e apresentado um estudo de setup para a construcao de um sistema nao-invasivo a ser implementado para diagnostico e prognostico de falhas em maquinas eletricas utilizando sinais sonoros captados por um microfone. Os resultados mostram as diferencas no espectro de frequencias entre motor saudavel, com adesbalanco eletrico, duas barras de rotor quebrado e corrosao nas pistas externas de rolamento. Os autores concluem a possibilidade de utilizacao de sinais sonoros para o diagnostico de falhas em motores eletricos.

Fault detection in electrical machines is a topic widely explored by researchers, especially bearing faults that represent about half of the total three-phase induction motor failure occurrences. This kind of fault is detectable by specific frequencies of the stator current and is a wide source of investigation. Thus, this work presents a predictability analysis method that provides patterns based on measures of relative entropy, Bhattacharyya distance, and Lempel–Ziv complexity estimated over reconstructed signals obtained from wavelet packet decomposition components. The signals under study were collected from motors with faults in the inner or outer races, which were artificially created in laboratory. These patterns were applied to three neural network topologies, which were used to classify the signals into two groups: normal or faulty.

Frequency domain analyses in electromyographic (EMG) signals are frequently applied to assess muscle fatigue and similar variables. Moreover, Fourier-based approaches are typically used for investigating these procedures. Nonetheless, Fourier analysis assumes the signal as stationary which is unlikely during dynamic contractions. As an alternative method, wavelet-based treatments do not assume this pattern and may be considered as more appropriate for joint time-frequency domain analysis. Based on the previous statements, the purpose of the present study was to compare the application of Short-Time Fourier Transform (STFT) and Continuous Wavelet Transform (CWT) to assess muscle fatigue in dynamic exercise of a 1-km of cycling (time-trial condition). The results of this study indicated that CWT and STFT analyses have provided similar fatigue estimates (slope) (p> 0.05). However, CWT application represents lesser dispersion (pp> 0.05) according to different methods, it is important to note that these responses seem to show greater values for CWT compared to STFT for 2 superficial muscles. Thereby, we are capable of considering CWT as a reliable and useful method to take into consideration when non-stationary or oscillating exercise models are evaluated.

In this work, we presented a methodology for extraction of patterns for failure classification of three-phase induction machines from their line currents. The signals were acquired from machines under four different conditions: normal operation, stator winding short circuit, rotor broken bars, and bearings faults. The method consists of projecting the signals into the dq axes by use of the Clarke and Park Transforms. The oscillations of these signals, that are due to power quality problems and the faulty conditions, are quantified by the standard deviation of the details components of a wavelet decomposition using several levels. The proposed patterns have visible differences for the machines faulty conditions. Thus, in order to test them, we used a Kohonen Self-Organized Map network to try to classify them obtaining high classification accuracy for machines working with 25% of their nominal torque.

Dynamic Time Warping (DTW), a pattern matching technique traditionally used for restricted vocabulary speech recognition, is based on a temporal alignment of the input signal with the template models. The principal drawback of DTW is its high computational cost as the lengths of the signals increase. This paper shows extended results over our previously published conference paper, which introduces an optimized version of the DTW I hat is based on the Discrete Wavelet Transform (DWT). (C) 2008 Elsevier B.V. All rights reserved.

Nowadays, noninvasive methods of diagnosis have increased due to demands of the population that requires fast, simple and painless exams. These methods have become possible because of the growth of technology that provides the necessary means of collecting and processing signals. New methods of analysis have been developed to understand the complexity of voice signals, such as nonlinear dynamics aiming at the exploration of voice signals dynamic nature. The purpose of this paper is to characterize healthy and pathological voice signals with the aid of relative entropy measures. Phase space reconstruction technique is also used as a way to select interesting regions of the signals. Three groups of samples were used, one from healthy individuals and the other two from people with nodule in the vocal fold and Reinke’s edema. All of them are recordings of sustained vowel /a/ from Brazilian Portuguese. The paper shows that nonlinear dynamical methods seem to be a suitable technique for voice signal analysis, due to the chaotic component of the human voice. Relative entropy is well suited due to its sensibility to uncertainties, since the pathologies are characterized by an increase in the signal complexity and unpredictability. The results showed that the pathological groups had higher entropy values in accordance with other vocal acoustic parameters presented. This suggests that these techniques may improve and complement the recent voice analysis methods available for clinicians.

This work presents a spoken document summarization (SDS) scheme that is based on an improved version of the Dynamic Time Warping (DTW) algorithm, and on the Discrete Wavelet Transform (DWT). Tests and results with sentences extracted from TIMIT speech corpus show the efficacy of the proposed technique.

This work describes a novel algorithm to identify laryngeal pathologies, by the digital analysis of the voice. It is based on Daubechies' discrete wavelet transform (DWT-db), linear prediction coefficients (LPC), and least squares support vector machines (LS-SVM). Wavelets with different support-sizes and three LS-SVM kernels are compared. Particularly, the proposed approach, implemented with modest computer requirements, leads to an adequate larynx pathology classifier to identify nodules in vocal folds. It presents over 90% of classification accuracy and has a low order of computational complexity in relation to the speech signal's length.

In this work, we present characteristics of an adaptive beamformer using Bayesian statistical inferences. We develop a model of array signal processing with optimum beamformer. The optimization of this beamformer focus on the uncertainty directions using a minimum mean-squared error spatial estimator. In spite of this, the underlying aim of this work is to present some issues about Bayesian implementations and generalized sidelobe cancelers. Also, we present some results of the beamformer a posteriori inference.

Ha uma necessidade de estudar formas de analise para sinais EMG, ja que este metodo nao constitui formas invasivas de captacao de sinais musculares. Este trabalho tem como foco principal a explicacao do funcionamento da analise de sinais EMG utilizando a Transformada Wavelet Continua, sendo esta apresentada como uma combinacao linear de uma mother Wavelet e seus respectivos coeficientes. Toda esta analise sera efetuada para que seja possivel verificar os valores da frequencia media do musculo, dividindo o sinal em pequenos pedacos no tempo para a averiguacao do decaimento da frequencia media, indicando assim, a fadiga muscular.

Redes de Sensores sem fio tem como objetivo monitorar algum fenomeno decorrente no ambiente, possuindo, assim, diversas aplicacoes. Este artigo apresenta a parte inicial do projeto que envolve uma RSSF projetada para monitoramento de temperatura de um determinado espaco atraves da modificacao do programa de teste apresentado em linguagem nesC proprio do software TinyOS. Devido a sua conversao de dados coletados analogicos para digitais, necessita-se de modificacao na implementacao do aplicativo Java para que seja codificada a temperatura em escala Celsius, facilitando a sua observacao. Diante disso, coletou-se em um determinado ambiente dados de temperatura na escala desejada.

Fault detection in electrical machines have been widely explored by researchers, especially bearing faults that represents about 40% to 60% of the total faults. Since this kind of fault is detectable by particular frequencies at the stator current, it is now a source of investigation. Thus, this work presents a predicability analysis method based on relative entropy measures estimated over reconstructed signals obtained from wavelet-packet decomposition components. The signals were simulated using a real motor current signal with addition of frequency components related to the bearing faults. Using three ANN topologies, these entropy measures are classified in two groups: normal and faulty signals with a high performance rate.

In this paper, we analyze voice signals recorded from patients with neurological disorders of different etiologies. The study was based on three samples of each patient: one before any ingestion, one after the swallowing of a liquid solution, and one after the swallowing of a pasty solution. We used three approaches: first, acoustical analysis, specifically fundamental frequency, jitter and shimmer; second, a proposed analysis method of vocal dynamic visual patterns, which are based on phase space reconstruction of the signals; and third, relative entropy analysis between the groups of signals. We show that the acoustical measures were not able to differentiate the study cases, relative entropy was only partially able to perform this task, but the visual patterns analysis was successful.

In this study, we present a first attempt to apply the predictable component analysis to voice signals. This method projects the signals in components that optimize the normalized error variance, a quantity related to the predictability of the signals. Using the proposed method along with a wavelet decomposition model, we obtain voice signals estimations. These voice signals are sustained vowel "a" from three groups of people: those with healthy larynxes and others with nodule or Reinke's edema on the vocal folds. Using the Shannon entropy of the estimation errors, we present evidences that it is possible to detect the pathological cases using this measure.

Este trabalho apresenta um estudo inicial da aplicação de medidas de previsibilidade em sinais de voz. O objetivo é desenvolver métodos que sejam capazes de diferenciar sinais saudáveis e patológicos, inclusive separar patologias. Para isso, tenta-se medir a variação da incerteza e previsibilidade entre os sinais de voz dos grupos analisados. Algumas patologias de laringe, como nódulo e edema de Reinke, usadas neste estudo, causam modificações nos sinais de voz devido a mudanças na estrutura e funcionalidade do trato e pregas vocais. Nos casos patológicos, tem-se, principalmente, aumento de perturbações de freqüência e amplitude, adição de ruído e supressão de componentes harmônicos de alta freqüência da voz. Por causa disso, observa-se perda da estrutura quase-periódica dos sinais, aumentando-se a incerteza do sistema vocal e, portanto, diminuição de sua previsibilidade. Para avaliar essas mudanças, usam-se medidas de entropia de Shannon e entropia relativa entre os sinais saudáveis e patológicos. Além dessas, tem-se a potência de previsão (PP), a qual é uma medida baseada na entropia relativa entre o sinal de voz e seu erro de previsão obtido por um modelo. Inicialmente, optou-se pelo modelo autorregressivo (AR), consagrado em análise de voz, porém, devido a resultados não satisfatórios, apresentou-se um modelo baseado em decomposições por wavelets. Outra ferramenta utilizada foi a chamada análise de componentes previsíveis (PrCA), a qual realiza uma decomposição dos sinais em componentes ordenados por sua previsibilidade, sendo possível reconstruí-los usando somente os componentes mais previsíveis. Também, com essa técnica, analisaram-se representações tridimensionais dos sinais de voz em um espaço cujas coordenadas são dadas por versões atrasadas dos próprios sinais. Os algoritmos desenvolvidos foram testados com o auxílio de sinais de voz simulados, os quais possuíam variações de nível de ruído e perturbações de amplitude e freqüência. Com isso, foi possível detectar erros e solucionar problemas com os métodos. Após a avaliação dos algoritmos, estimou-se os valores de entropia dos sinais de voz, a entropia relativa entre os sinais saudáveis e os sinais dos grupos analisados, além de se calcular a PP usando o modelo AR e o modelo por wavelets. Por fim, utilizou-se a PrCA para obtenção de versões mais previsíveis dos sinais, então, calculando-se a PP para esses casos usando essa versão como previsão dos sinais. Aplicou-se, também, a PrCA para as representações tridimensionais dos sinais usando um modelamento AR multidimensional para obtenção de previsões. Com os ensaios de entropia dos sinais de voz, não foi possível diferenciar os grupos, mas com os resultados de entropia relativa, conseguiu-se distinguir eficientemente os sinais patológicos dos saudáveis. Porém, essa medida não possui muita aplicação prática, isso pois é necessário um banco de vozes diagnosticadas para servir de comparação. Nos ensaios de PP usando modelo AR, também não foi possível diferenciar os grupos, no entanto, com o modelo wavelet, os sinais saudáveis apresentaram significativamente maior previsibilidade do que os patológicos, mas, mesmo assim, não se conseguiu diferenciar as patologias. Com a PrCA, utilizando-se ambos os modelos, foi possível diferenciar os grupos patológicos do saudável, porém, frente ao modelo AR, os sinais saudáveis apresentaram menor previsibilidade. Isso demonstra que a previsibilidade depende do modelo usado para a análise, assim, as patologias da laringe podem diminuir ou aumentar a capacidade de previsão dos sinais de voz conforme o modelo usado. Com a avaliação dos resultados de PrCA das representações tridimensionais, tem-se comportamento semelhante ao obtido pela análise direta nos sinais de voz com o modelo AR, entretanto, essa forma e representação dos dados mostra se promissora em estudos futuros. Com esses ensaios, concluiu-se que este estudo foi muito útil para um maior conhecimento da dinâmica da produção vocal e que as medidas de previsibilidade são interessantes para avaliação de patologias da laringe, em especial, a presença de nódulo nas pregas vocais e edema de Reinke, pelo menos nestes estudos iniciais e usando os sinais de voz disponíveis. Mais estudos ainda são necessários, entretanto essa forma de análise já apresenta bons resultados, os quais podem ser aplicados para auxiliar o diagnóstico de disfonias por profissionais da saúde. This thesis presents initial studies of the application of predictability measures to voice signal analysis. Its aim is to develop methods that are capable of differentiating healthy and pathological signals, also amongst pathologies. In order to do that, we perform an attempt to measure the uncertainty and predictability variations of the signals from the analyzed groups. Some larynx pathologies, such as nodule and Reinkes edema, that are used in this study, cause changes to the voice signals due to structure and functionality modifications of the vocal tract and folds. The main modifications are higher amplitude and frequency perturbations, noise addition, and supression of high frequency harmonic components. Because of that, the signals lose some of their almost periodic structure, the vocal system\'s uncertainty increases and, therefore, the predictability decreases. We use several measures to evaluate these changes, such as Shannons entropy and relative entropy between healthy and pathological signals. In addition, we use the predictive power (PP), that is based on the relative entropy between the voice signal and its prediction error given by a model. Firstly, we used the autoregressive model (AR), common for voice analysis, however, due to unsatisfactory results, we presented a model based on wavelet decomposition. We also took advantage of another tool, called predictable component analysis (PrCA), it performs a signal decomposition in components that are ordered by their predictability. Then it is possible to reconstruct the signals using only their most predictable components. Using this technique, we analyzed a kind of tridimensional representation of the voice signals in a space with coordinates given by delayed versions of the signals. We tested the developed algorithms with the aid of simulated voice signals, which had variations of noise level and amplitude and frequency perturbations. By means of that, it was possible to detect errors and solve method problems. After the algorithms evaluation, we estimated the entropy of the voice signals and the relative entropy between the healthy signals and all the signals. In addition, we estimated the PP using the AR and wavelet based models. After that, we used the PrCA in order to obtain more predictable versions of the signals and then, estimated the PP using this version as the signals prediction. Also, we applied the PrCA to the signals tridimensional representations using a multidimensional AR model as a predictor. Using the voice entropy results, we could not distinguish between the analyzed groups, but with the relative entropy values, the healthy and pathological signals were differentiated efficiently. In spite of that, this measure has no practical application, because a diagnosed voice database is necessary as a basis of comparison. For the PP with AR modeling, no distinction between the groups is observed, but with the wavelet modeling, the healthy signals showed significantly higher predictability than the pathological ones, however the pathologies were differentiated. Using the PrCA with both models, the pathological and healthy groups were distinguished, but for the AR model, the healthy signals presented smaller predictability. This shows that the predictability depends on the analysis model, thus the larynx pathologies can decrease or increase the prediction capacity of the voice signals according to the used model. The results of PrCA of the tridimensional representations show similar behavior of the ones from direct PrCA signal analisys with the AR model. Despite of these results, this form of data representation seems to be promising for future studies. Considering these results, we concluded that this study was very useful to acquire a better understanding of the dynamics of voice production and that the predictability measures are interesting for the evaluation of larynx pathologies, especially presence of nodule in the vocal folds and Reinke\'s edema, at least for this initial study using the available signals. More studies are still necessary, but this analysis method already presents good results, which can be applied to aid pathology diagnosis by health professionals.

This paper, proposes a method of voice signal analysis based on predictability measures. The objective is the differentiation of three groups of signals: healthy, nodules on the vocal folds, and Reinke's edema. This method is a slight reformulation of a measure called predictive power (PP) that uses the relative entropy between the signal and its prediction error to capture information about the predictability of the patient's voice production system. A wavelet decomposition model provides the predictions. The results obtained for the healthy group were substantially higher than those of the pathologic groups. This decrease of predictability may be interpreted as an indication of the increase of intricacy and irregularity associated with the studied larynx pathologies.

In this paper we present a new wavelet-based algorithm for low-cost computation of the cepstrum. It can be used for real time precise pitch determination in automatic speech and speaker recognition systems. Many wavelet families are examined to determine the one that works best. The results confirm the efficacy and accuracy of the proposed technique for pitch extraction. (C) 2008 Elsevier B.V. All rights reserved.

Nowadays, fractal analysis has been successfully applied to digital speech processing, particularly for word and phoneme segmentation, which represents one of the fundamental steps in automatic speech recognition systems. The practical use of fractal analysis for this purpose should match two principles: low computational cost, to allow the use in real-time, and accuracy in the results, in order to produce a satisfactory segmentation, sending the correct data to the classifier. Aiming at meeting these two requirements, this work proposes a technique for speech segmentation based on the fractal dimension, which is obtained by using the discrete wavelet transform that avoids the use of 1/k pre-filtering. Many families of wavelets are presented and compared, and the results assure the efficacy of the proposed method.

In the last years the use of voice acoustical analysis has increased much. Also, other techniques are being applied, such as nonlinear methods and comparison of quantities obtained from information theory. The latter is the case of this paper, we analyze characteristics of the entropy rate curve and its asymptote to acquire knowledge about the evolution of pathological voice signals. Specifically signals from patients with nodule in the vocal folds. The analysis provided some differences between nodule and healthy signals that may be related to the level of evolution of the pathology.

This paper introduces the paradigm of joint time- frequency-shape analysis of discrete-time signals by means of a novel transform called discrete shapelet transform (DST). This approach allows one to And the time-support of frequencies, and, at the same time, analyze a shape. Applications are possible for a diversity of fields in signal processing, and the results assure the efficacy of the proposed transform.

This letter describes a novel algorithm that is based on autoregressive decomposition and pole tracking used to recognize two patterns of speech data: normal voice and disphonic voice caused by nodules. The presented method relates the poles and the peaks of the signal spectrum which represent the periodic components of the voice. The results show that the perturbation contained in the signal is clearly depicted by pole's positions. Their variability is related to jitter and shimmer. The pole dispersion for pathological voices is about 20% higher than for normal voices, therefore, the proposed approach is a more trustworthy measure than the classical ones. (c) 2007 Published by Elsevier B.V.

In this paper, it is presented an insight of the effects of the application of autoregressive (AR) decomposition and pole tracking to voice signals. The AR model is used to decompose the signals in a set of poles which has a correspondence to the peaks of the signals power spectral density function (PSD). The aim of this work is to show the differences in the behavior of these poles for voice signals collected from two groups of people, one with healthy glottal tract and another with nodule pathology in vocal folds.