No description available.

[pt] Esta tese apresenta e examina contribuições inovadoras no front-end dos sistemas de reconhecimento automático de voz (RAV) para o realce e reconhecimento de voz em ambientes adversos. A primeira proposta consiste em aplicar um filtro de mediana sobre a função de distribuição de probabilidade de cada coeficiente cepstral antes de utilizar uma transformação para um domínio invariante às distorções, com o objetivo de adaptar a voz ruidosa ao ambiente limpo de referência através da modificação de histogramas. Fundamentadas nos resultados de estudos psicofísicos do sistema auditivo humano, que utiliza como princípio o fato de que o som que atinge o ouvido é sujeito a um processo chamado Análise de Cena Auditiva (ASA), o qual examina como o sistema auditivo separa as fontes de som que compõem a entrada acústica, três novas abordagens aplicadas independentemente foram propostas para realce e reconhecimento de voz. A primeira aplica a estimativa de uma nova máscara no domínio espectral usando o conceito da transformada de Fourier de tempo curto (STFT). A máscara proposta aplica a técnica Local Binary Pattern (LBP) à relação sinal ruído (SNR) de cada unidade de tempo-frequência (T-F) para estimar uma máscara de vizinhança ideal (INM). Continuando com essa abordagem, propõe-se em seguida nesta tese o mascaramento usando as transformadas wavelet com base nos LBP para realçar os espectros temporais dos coeficientes wavelet nas altas frequências. Finalmente, é proposto um novo método de estimação da máscara INM, utilizando um algoritmo de aprendizagem supervisionado das Deep Neural Networks (DNN) com o objetivo de realizar a classificação de unidades T-F obtidas da saída dos bancos de filtros pertencentes a uma mesma fonte de som (ou predominantemente voz ou predominantemente ruído). O desempenho é comparado com as técnicas de máscara tradicionais IBM e IRM, tanto em termos de qualidade objetiva da voz, como através de taxas de erro de palavra. Os resultados das técnicas propostas evidenciam as melhoras obtidas em ambientes ruidosos, com diferenças significativamente superiores às abordagens convencionais. / [en] This thesis presents and examines innovative contributions in frontend of the automatic speech recognition systems (ASR) for enhancement and speech recognition in adverse environments. The first proposal applies a median filter on the probability distribution function of each cepstral coefficient before using a transformation to a distortion-invariant domain, to adapt the corrupted voice to the clean reference environment by modifying histograms. Based on the results of psychophysical studies of the human auditory system, which uses as a principle the fact that sound reaching the ear is subjected to a process called Auditory Scene Analysis (ASA), which examines how the auditory system separates the sound sources that make up the acoustic input, three new approaches independently applied were proposed for enhancement and speech recognition. The first applies the estimation of a new mask in the spectral domain using the short-time Fourier Transform (STFT) concept. The proposed mask applies the Local Binary Pattern (LBP) technique to the Signal-to-Noise Ratio (SNR) of each time-frequency unit (T-F) to estimate an Ideal Neighborhood Mask (INM). Continuing with this approach, the masking using LBP-based wavelet transforms to highlight the temporal spectra of wavelet coefficients at high frequencies is proposed in this thesis. Finally, a new method of estimation of the INM mask is proposed, using a supervised learning algorithm of Deep Neural Network (DNN) to classify the T-F units obtained from the output of the filter banks belonging to a same source of sound (or predominantly voice or predominantly noise). The performance is compared with traditional IBM and IRM mask techniques, both regarding objective voice quality and through word error rates. The results of the proposed methods show the improvements obtained in noisy environments, with differences significantly superior to the conventional approaches.

No description available.

[pt] Nesta dissertação de mestrado, nós desenvolvemos um processo de litografia baseado na técnica de microscopia de força atômica. O estudo do processo de litografia aqui utilizado inicia-se com a deposição e caracterização de filmes finos de sulfeto de arsênio amorfo (a-As2S3) em substratos de silício e a deposição de uma camada metálica de alumínio, utilizada como máscara, sobre a superfície do a-As2S3. O microscópio de força atômica é utilizado para escrever os padrões de forma controlada na camada metálica, e para tal, a influencia dos parâmetros de controle do microscópio na realização da litografia foi analisada. Para a transferência do padrão litografado realiza-se um posterior processo de fotossensibilização e dissolução química do a-As2S3 com uma solução de K2CO3. Após a dissolução, uma camada de ouro foi depositada por erosão catódica DC, seguido de uma nova dissolução, desta vez com NaOH resultando na transferência de nanoestruturas de Au para o substrato de silício. / [en] In this dissertation, we have developed a lithography process based on the atomic force microscopy of technique. The study of the lithography process starts with the deposition and characterization of amorphous arsenic sulfide thin films (a-As2S3) in silicon substrates and the deposition of a metallic aluminum layer, used as mask, on the surface of the a-As2S3. An atomic force microscope was used to write patterns in a controlled way on the metallic layer. Therefore, the influence of microscope feedback system on the accomplishment of the lithography was analyzed. In order to transfer the lithographed pattern to a silicon substrate, the a- As2S3 was exposed to a UV light source and was dissolved with a K2CO3 solution. Then, a thin gold layer was deposited by sputtering DC, and a new dissolution, now with NaOH was performed, leading to the deposition of Au nanostructures onto the silicon substrate.

This investigation into modern Cuban theater examines the self-reflective discourse of Afro-Cuban women and her masks in the mirror. This study arose in response to the lack of critical analysis undertaken on this subject matter. Although the number of plays with female Afro-Cuban protagonists is relatively small, there is a growing body of plays with specific concepts that identify them as Contemporary Afro-Cuban Theater, where the life and culture of the Afro-Cuban people is viewed as through a mirror commenting on their own experiences. The persistence of certain themes, formal elements, staging, and an anti-colonial world view form the unique markers of this genre within Cuban theater. The three plays that constitute the primary material for this study are: Juan Revolico (1944), by Flora Diaz Parrado, Medea en el espejo (1959), by Jose Triana, and Maria Antonia (1964), by Eugenio Espinosa Hernandez. Each of these plays is a tragedy with a modern heroin that challenges the stereotypes established by a patriarchal and Eurocentric society, and creating new archetypes.The investigation analyzes the disobedience and rebelliousness of the heroines as they strive against the established power structure. It uncovers a unique approach to Cuban national identity in which the marginalized Afro-Cuban woman is cast in a protagonist role; it is through these protagonist roles that she subverts pre-established notions of race, class, and gender, from the women's tragedies in the patriarchal Euro-centric Cuban society.A re-structuring of gender roles is brought about through the feminine powers of seduction. Throughout each play the Afro-Cuban woman sees herself in the mirror and is able to gather the pieces of her fragmented identity. Afro-Cuban culture and religion are central to the solution of the thematic, formal, and plot conflicts.Among the most relevant contributions of this study is the idea that contemporary Afro-Cuban Theater exists as its own genre. Afro-Cuban Theater exists with its own aesthetics, profile, and anti-colonial perspective. Here the perspectives of Afro-Cuban woman are explored through the protagonists of three of the most significant models of this genre, Juan Revolico, Medea en el espejo, and Maria Antonia.

Les problèmes majeurs de la génération automatique du dessin des masques des circuits intégrés sont la dépendance vis-à-vis des règles de dessin et le dimensionnement correct des transistors. Les méthodes traditionnelles, telles que l'utilisation de cellules pré-caractérisées, manquent de flexibilité, car les portes des bibliothèques (en nombre limité) sont dessinées et dimensionnées (independarnment de l'application) pour une technologie donnée. Les méthodes de synthèse automatique du dessin des masques ont pour but de surmonter ces problèmes. Les techniques les plus couramment utilisées sont le "gate-matrix" et le "linear-matrix". L'indépendance vis-à-vis des règles de dessin est obtenue en utilisant la technique de description symbolique (dessin sous une grille unitaire), et les dimensions des transistors sont définies par le concepteur ou par un outil de dimensionnement. Nous proposons une méthode et un prototype logiciel pour la synthèse automatique des masques, en utilisant le style "linear-matrix multi-bander". La description d'entree du générateur est un fichier format SPICE (au niveau transistor), ce qui permet d'avoir un nombre très élevé de cellules, en particulier les portes complexes (A01), et ainsi avoir une meilleure optimisation lors de la phase d'assignation technologique. Les macro-cellules générées doivent être assemblées afin de réaliser un circuit complet. Deux contraintes supplémentaires sont ainsi imposées au générateur: malléabilité de la forme et position des broches d'entrées/sorties sur la périphérie de la macro-cellule. Les macro-cellules sont assemblées en utilisant un environnement de conception industriel. Les contributions de ce mémoire de doctorat sont d'une part le développement d'un générateur de macro-cellules flexible ayant les caracteristiques d'indépendance aux règles de dessin et d'intégration dans un environnement de macro-cellules, et d'autre part l'étude detailée des paramètres qui déterminent la surface occupée, les performances électriques et la puissance dissipée des macro-cellules générées automatiquement. / The main problems of the automatic layout synthesis are the design rules dependence and the transistor sizing. The traditional layout synthesis methods, like standard-cells, are not flexible, since the cells in the libraries are designed and sized for a specific technology. In this way, the designer must change his library at each technology improvement. The automatic layout synthesis methods overcomes these problems (design rules dependence and transistor sizing). Examples of layout styles are gate-matrix and linear-matrix. The technology independence is achieved by symbolic description (layout under an unitary grid), and the transistor sizes are defined by the designer or by a sizing tool. From these two constraints, we develop an automatic layout synthesis tool, using a linear-matrix multi-row layout style. The input description for our tool is a Spice file. This descriptions allows to define a greater number of cells (mainly AOIs gates), resulting a technology mapping with less constraints. The generated macro-cells must be assembled in order to construct a complete circuit. Two additional constraints are then imposed to the generator : variable aspect ratio and placement of the inputs/outputs pins in the macro-cell border. The macro-cells are assembled by an industrial CAD environment. The main contributions of this thesis are the development of a macro-cell generator (with the characteristics of technology independence and easy integration in a macro-cell environment) and the analysis of the parameters playing a role in the area, delay and power consumption.

Les problèmes majeurs de la génération automatique du dessin des masques des circuits intégrés sont la dépendance vis-à-vis des règles de dessin et le dimensionnement correct des transistors. Les méthodes traditionnelles, telles que l'utilisation de cellules pré-caractérisées, manquent de flexibilité, car les portes des bibliothèques (en nombre limité) sont dessinées et dimensionnées (independarnment de l'application) pour une technologie donnée. Les méthodes de synthèse automatique du dessin des masques ont pour but de surmonter ces problèmes. Les techniques les plus couramment utilisées sont le "gate-matrix" et le "linear-matrix". L'indépendance vis-à-vis des règles de dessin est obtenue en utilisant la technique de description symbolique (dessin sous une grille unitaire), et les dimensions des transistors sont définies par le concepteur ou par un outil de dimensionnement. Nous proposons une méthode et un prototype logiciel pour la synthèse automatique des masques, en utilisant le style "linear-matrix multi-bander". La description d'entree du générateur est un fichier format SPICE (au niveau transistor), ce qui permet d'avoir un nombre très élevé de cellules, en particulier les portes complexes (A01), et ainsi avoir une meilleure optimisation lors de la phase d'assignation technologique. Les macro-cellules générées doivent être assemblées afin de réaliser un circuit complet. Deux contraintes supplémentaires sont ainsi imposées au générateur: malléabilité de la forme et position des broches d'entrées/sorties sur la périphérie de la macro-cellule. Les macro-cellules sont assemblées en utilisant un environnement de conception industriel. Les contributions de ce mémoire de doctorat sont d'une part le développement d'un générateur de macro-cellules flexible ayant les caracteristiques d'indépendance aux règles de dessin et d'intégration dans un environnement de macro-cellules, et d'autre part l'étude detailée des paramètres qui déterminent la surface occupée, les performances électriques et la puissance dissipée des macro-cellules générées automatiquement. / The main problems of the automatic layout synthesis are the design rules dependence and the transistor sizing. The traditional layout synthesis methods, like standard-cells, are not flexible, since the cells in the libraries are designed and sized for a specific technology. In this way, the designer must change his library at each technology improvement. The automatic layout synthesis methods overcomes these problems (design rules dependence and transistor sizing). Examples of layout styles are gate-matrix and linear-matrix. The technology independence is achieved by symbolic description (layout under an unitary grid), and the transistor sizes are defined by the designer or by a sizing tool. From these two constraints, we develop an automatic layout synthesis tool, using a linear-matrix multi-row layout style. The input description for our tool is a Spice file. This descriptions allows to define a greater number of cells (mainly AOIs gates), resulting a technology mapping with less constraints. The generated macro-cells must be assembled in order to construct a complete circuit. Two additional constraints are then imposed to the generator : variable aspect ratio and placement of the inputs/outputs pins in the macro-cell border. The macro-cells are assembled by an industrial CAD environment. The main contributions of this thesis are the development of a macro-cell generator (with the characteristics of technology independence and easy integration in a macro-cell environment) and the analysis of the parameters playing a role in the area, delay and power consumption.

Les problèmes majeurs de la génération automatique du dessin des masques des circuits intégrés sont la dépendance vis-à-vis des règles de dessin et le dimensionnement correct des transistors. Les méthodes traditionnelles, telles que l'utilisation de cellules pré-caractérisées, manquent de flexibilité, car les portes des bibliothèques (en nombre limité) sont dessinées et dimensionnées (independarnment de l'application) pour une technologie donnée. Les méthodes de synthèse automatique du dessin des masques ont pour but de surmonter ces problèmes. Les techniques les plus couramment utilisées sont le "gate-matrix" et le "linear-matrix". L'indépendance vis-à-vis des règles de dessin est obtenue en utilisant la technique de description symbolique (dessin sous une grille unitaire), et les dimensions des transistors sont définies par le concepteur ou par un outil de dimensionnement. Nous proposons une méthode et un prototype logiciel pour la synthèse automatique des masques, en utilisant le style "linear-matrix multi-bander". La description d'entree du générateur est un fichier format SPICE (au niveau transistor), ce qui permet d'avoir un nombre très élevé de cellules, en particulier les portes complexes (A01), et ainsi avoir une meilleure optimisation lors de la phase d'assignation technologique. Les macro-cellules générées doivent être assemblées afin de réaliser un circuit complet. Deux contraintes supplémentaires sont ainsi imposées au générateur: malléabilité de la forme et position des broches d'entrées/sorties sur la périphérie de la macro-cellule. Les macro-cellules sont assemblées en utilisant un environnement de conception industriel. Les contributions de ce mémoire de doctorat sont d'une part le développement d'un générateur de macro-cellules flexible ayant les caracteristiques d'indépendance aux règles de dessin et d'intégration dans un environnement de macro-cellules, et d'autre part l'étude detailée des paramètres qui déterminent la surface occupée, les performances électriques et la puissance dissipée des macro-cellules générées automatiquement. / The main problems of the automatic layout synthesis are the design rules dependence and the transistor sizing. The traditional layout synthesis methods, like standard-cells, are not flexible, since the cells in the libraries are designed and sized for a specific technology. In this way, the designer must change his library at each technology improvement. The automatic layout synthesis methods overcomes these problems (design rules dependence and transistor sizing). Examples of layout styles are gate-matrix and linear-matrix. The technology independence is achieved by symbolic description (layout under an unitary grid), and the transistor sizes are defined by the designer or by a sizing tool. From these two constraints, we develop an automatic layout synthesis tool, using a linear-matrix multi-row layout style. The input description for our tool is a Spice file. This descriptions allows to define a greater number of cells (mainly AOIs gates), resulting a technology mapping with less constraints. The generated macro-cells must be assembled in order to construct a complete circuit. Two additional constraints are then imposed to the generator : variable aspect ratio and placement of the inputs/outputs pins in the macro-cell border. The macro-cells are assembled by an industrial CAD environment. The main contributions of this thesis are the development of a macro-cell generator (with the characteristics of technology independence and easy integration in a macro-cell environment) and the analysis of the parameters playing a role in the area, delay and power consumption.