Bei der Überführung eines wissenschaftlichen Laborsystems zur automatischen Spracherkennung in eine reale Anwendung ergeben sich verschiedene praktische Problemstellungen, von denen eine der Verlust an Erkennungsleistung durch umgebende akustische Störungen ist. Im Gegensatz zu additiven Störungen wie Lüfterrauschen o. ä. hat die Wissenschaft bislang die Störung des Raumhalls bei der Spracherkennung nahezu ignoriert. Dabei besitzen, wie in der vorliegenden Dissertation deutlich gezeigt wird, bereits geringfügig hallende Räume einen stark störenden Einfluss auf die Leistungsfähigkeit von Spracherkennern. Mit dem Ziel, die Erkennungsleistung wieder in einen praktisch benutzbaren Bereich zu bringen, nimmt sich die Arbeit dieser Problemstellung an und schlägt Lösungen vor. Der Hintergrund der wissenschaftlichen Aktivitäten ist die Erstellung von funktionsfähigen Sprachbenutzerinterfaces für Gerätesteuerungen im Wohn- und Büroumfeld, wie z.~B. bei der Hausautomation. Aus diesem Grund werden praktische Randbedingungen wie die Restriktionen von embedded Computerplattformen in die Lösungsfindung einbezogen. Die Argumentation beginnt bei der Beschreibung der raumakustischen Umgebung und der Ausbreitung von Schallfeldern in Räumen. Es wird theoretisch gezeigt, dass die Störung eines Sprachsignals durch Hall von zwei Parametern abhängig ist: der Sprecher-Mikrofon-Distanz (SMD) und der Nachhallzeit T60. Um die Abhängigkeit der Erkennungsleistung vom Grad der Hallstörung zu ermitteln, wird eine Anzahl von Erkennungsexperimenten durchgeführt, die den Einfluss von T60 und SMD nachweisen. Weitere Experimente zeigen, dass die Spracherkennung kaum durch hochfrequente Hallanteile beeinträchtigt wird, wohl aber durch tieffrequente. In einer Literaturrecherche wird ein Überblick über den Stand der Technik zu Maßnahmen gegeben, die den störenden Einfluss des Halls unterdrücken bzw. kompensieren können. Jedoch wird auch gezeigt, dass, obwohl bei einigen Maßnahmen von Verbesserungen berichtet wird, keiner der gefundenen Ansätze den o. a. praktischen Einsatzbedingungen genügt. In dieser Arbeit wird die Methode Harmonicity-based Feature Analysis (HFA) vorgeschlagen. Sie basiert auf drei Ideen, die aus den Betrachtungen der vorangehenden Kapitel abgeleitet werden. Experimentelle Ergebnisse weisen die Verbesserung der Erkennungsleistung in halligen Umgebungen nach. Es werden sogar praktisch relevante Erkennungsraten erzielt, wenn die Methode mit verhalltem Training kombiniert wird. Die HFA wird gegen Ansätze aus der Literatur evaluiert, die ebenfalls praktischen Implementierungskriterien genügen. Auch Kombinationen der HFA und einigen dieser Ansätze werden getestet. Im letzten Kapitel werden die beiden Basistechnologien Stimm\-haft-Stimmlos-Entscheidung und Grundfrequenzdetektion umfangreich unter Hallbedingungen getestet, da sie Voraussetzung für die Funktionsfähigkeit der HFA sind. Als Ergebnis wird dargestellt, dass derzeit für beide Technologien kein Verfahren existiert, das unter Hallbedingungen robust arbeitet. Es kann allerdings gezeigt werden, dass die HFA trotz der Unsicherheiten der Verfahren arbeitet und signifikante Steigerungen der Erkennungsleistung erreicht. / Automatic speech recognition (ASR) systems used in real-world indoor scenarios suffer from performance degradation if noise and reverberation conditions differ from the training conditions of the recognizer. This thesis deals with the problem of room reverberation as a cause of distortion in ASR systems. The background of this research is the design of practical command and control applications, such as a voice controlled light switch in rooms or similar applications. Therefore, the design aims to incorporate several restricting working conditions for the recognizer and still achieve a high level of robustness. One of those design restrictions is the minimisation of computational complexity to allow the practical implementation on an embedded processor. One chapter comprehensively describes the room acoustic environment, including the behavior of the sound field in rooms. It addresses the speaker room microphone (SRM) system which is expressed in the time domain as the room impulse response (RIR). The convolution of the RIR with the clean speech signal yields the reverberant signal at the microphone. A thorough analysis proposes that the degree of the distortion caused by reverberation is dependent on two parameters, the reverberation time T60 and the speaker-to-microphone distance (SMD). To evaluate the dependency of the recognition rate on the degree of distortion, a number of experiments has been successfully conducted, confirming the above mentioned dependency of the two parameters, T60 and SMD. Further experiments have shown that ASR is barely affected by high-frequency reverberation, whereas low frequency reverberation has a detrimental effect on the recognition rate. A literature survey concludes that, although several approaches exist which claim significant improvements, none of them fulfils the above mentioned practical implementation criteria. Within this thesis, a new approach entitled 'harmonicity-based feature analysis' (HFA) is proposed. It is based on three ideas that are derived in former chapters. Experimental results prove that HFA is able to enhance the recognition rate in reverberant environments. Even practical applicable results are achieved when HFA is combined with reverberant training. The method is further evaluated against three other approaches from the literature. Also combinations of methods are tested. In a last chapter the two base technologies fundamental frequency (F0) estimation and voiced unvoiced decision (VUD) are evaluated in reverberant environments, since they are necessary to run HFA. This evaluation aims to find one optimal method for each of these technologies. The results show that all F0 estimation methods and also the VUD methods have a strong decreasing performance in reverberant environments. Nevertheless it is shown that HFA is able to deal with uncertainties of these base technologies as such that the recognition performance still improves.

This study evaluates an automated formant estimation procedure designed to adapt to speakers and variations in speech. The adaption is achieved by using the formant ceiling with the least variation (in combined estimates of F1 and F2) as the optimal ceiling. This optimization renders the best possible estimations given the data, therefore it could presumably also adapt to variations such as high fo. The procedure has not been evaluated by using material with known formant frequencies. Therefore, this is done here. The performance of the procedure is tested through comparison with a common procedure with fixed ceilings, based on speaker sex. The estimations are carried out on synthetic vowel tokens, systematically varied in formant frequencies and in fo, to match the natural variation within vowels and between speakers. The formant estimations are compared to target values, compared between procedures and to earlier studies. The results reveal that the formant estimation procedure with optimized ceilings does not perform better than the common procedure. Both procedures perform better than earlier methods, but neither deals satisfactorily with high fo. / Denna studie utvärderar en automatisk formantmätningsprocedur utvecklad för anpassning efter talare och variationer i tal. Anpassningen åstadkoms genom att använda det formanttak som uppvisar minst variation (i mätningar av F1 och F2 i kombination) som det optimerade taket. Denna optimering ger bästa möjliga estimeringar utifrån data, därför skulle troligtvis anpassningen även kunna ske till variation såsom hög fo. Proceduren har inte utvärderats genom att använda material med kända formantfrekvenser, varför det görs här. Formantmätningsprocedurens prestation testas genom jämförelse med gängse procedur med fasta formanttak, baserade på skillnader mellan kön. Formantmätningarna utförs på syntetiska vokalexemplar, systematiskt varierade i formantfrekvenser och i fo för att motsvara naturlig variation inom vokaler och mellan talare. Formantmätningarna jämförs mot ursprungsvärdena, procedurerna sinsemellan och med tidigare studier. Resultatet visar att formantmätningsproceduren med optimerat formanttak inte presterar bättre än den gängse proceduren. Båda procedurer presterar bättre än tidigare metoder, men ingen hanterar hög fo på ett tillfredställande sätt.

No description available.

No description available.

No description available.

No description available.

No description available.

Tesis por compendio / [EN] Different electronic interfaces have been proposed to measure major parameters for the characterization of quartz crystal microbalance (QCM) during the last two decades. The measurement of the adequate parameters of the sensor for a specific application is very important, since an error in this measure can lead to an error in the interpretation of the results. The requirements of the system of characterization depend on the application. In this thesis we propose two characterization systems for two types of applications that involve the majority of sensor applications: 1) Characterization of materials under variable damping conditions and 2) Detection of substances with high measurement resolution. The proposed systems seek to solve the problems detected in the systems currently in use. For applications in which the sensor damping varies during the experiment, we propose a system based on a new configuration of the technique of automatic capacitance compensation (ACC). This new configuration provides the measure of the series resonance frequency, the motional resistance and the parallel capacitance of the sensor. Moreover, it allows an easy calibration of the system that improves the precision in the measurement. We show the experimental results for 9 and 10 MHz crystals in fluid media, with different capacitances in parallel, showing the effectiveness in the capacitance compensation. The system presents some deviation in frequency with respect to the series resonance frequency, as measured with an impedance analyser. These deviations are due to the non-ideal, specific behaviour of some of the components of the circuit. A new circuit is proposed as a possible solution to this problem. For high-resolution applications we propose an integrated platform to characterize high-frequency acoustic sensors. The proposed system is based on a new concept in which the sensor is interrogated by means of a very stable, low-noise external source at a constant frequency, while the changes provoked by the charge in the phase of the sensor are monitored. The use of high-frequency sensors enhances the sensitivity of the measure, whereas the design characterization system reduces the noise in the measurement. The result is an improvement in the limit of detection (LOD). This way, we achieve one of the challenges in the acoustic high-frequency devices. The validation of the platform is performed by means of an immunosensor based in high fundamental frequency QCM crystals (HFF-QCM) for the detection of two pesticides: carbaryl and thiabendazole. The results obtained for carbaryl are compared to the results obtained by another high-frequency acoustic technology based in Love sensors, with the optical technique based in surface plasmonic resonance and with the gold standard technique Enzyme Linked Immunoassay (ELISA). The LOD obtained with the acoustic sensors HFF-QCM and Love is similar to the one obtained with ELISA and improves by one order of magnitude the LOD obtained with SPR. The conceptual ease of the proposed system, its low cost and the possibility of miniaturization of the quartz resonator, allows the characterization of multiple sensors integrated in an array configuration, which will allow in the future to achieve the challenge of multianalyte detection for applications of High-Throughput Screening (HTS). / [ES] Durante las dos últimas décadas se han propuesto diferentes interfaces electrónicos para medir los parámetros más importantes de caracterización de los cristales de microbalanza de cuarzo (QCM). La medida de los parámetros adecuados del sensor para una aplicación específica es muy importante, ya que un error en la medida de dichos parámetros puede resultar en un error en la interpretación de los resultados. Los requerimientos del sistema de caracterización dependen de la aplicación. En esta tesis se proponen dos sistemas de caracterización para dos ámbitos de aplicación que comprenden la mayoría de las aplicaciones con sensores QCM: 1) Caracterización de materiales bajo condiciones de amortiguamiento variable y 2) detección de sustancias con alta resolución de medida. Los sistemas propuestos tratan de resolver la problemática detectada en los ya existentes. Para aplicaciones en las que el amortiguamiento del sensor varía durante el experimento, se propone un sistema basado en una nueva configuración de la técnica de compensación automática de capacidad (ACC). La nueva configuración proporciona la medida de la frecuencia de resonancia serie, la resistencia dinámica y la capacidad paralelo del sensor. Además, permite una fácil calibración del sistema que mejora la precisión en la medida. Se presentan resultados experimentales para cristales de 9 y 10MHz en medios fluidos, con diferentes capacidades en paralelo, demostrando la efectividad de la compensación de capacidad. El sistema presenta alguna desviación en frecuencia con respecto a la frecuencia resonancia serie, medida con un analizador de impedancias. Estas desviaciones son explicadas convenientemente, debidas al comportamiento no ideal específico de algunoscomponentes del circuito. Una nueva propuesta de circuito se presenta como posible solución a este problema. Para aplicaciones de alta resolución se propone una plataforma integrada para caracterizar sensores acústicos de alta frecuencia. El sistema propuesto se basa en un nuevo concepto en el que el sensor es interrogado, mediante una fuente externa muy estable y de muy bajo ruido, a una frecuencia constante mientras se monitorizan los cambios producidos por la carga en la fase del sensor. El uso de sensores de alta frecuencia aumenta la sensibilidad de la medida, por otro lado, el sistema de caracterización diseñado reduce el ruido en la misma. El resultado es una mejora del límite de detección (LOD). Se consigue con ello uno de los retos pendientes en los dispositivos acústicos de alta frecuencia. La validación de la plataforma desarrollada se realiza con una aplicación de un inmunosensor basado en cristales QCM de alta frecuencia fundamental (HFF-QCM) para la detección de dos pesticidas: carbaryl y tiabendazol. Los resultados obtenidos para el Carbaryl se comparan con los obtenidos con otra tecnología acústica de alta frecuencia basada en sensores Love, con la técnica óptica basada resonancia superficial de plasmones (SPR) y con la técnica de referencia Enzyme Linked Immuno Assay (ELISA). El LOD obtenido con los sensores acústicos HFFQCM y Love es similar al obtenido con las técnicas ELISA y mejora en un orden de magnitud al obtenido con SPR. La sencillez conceptual del sistema propuesto junto con su bajo coste, así como la capacidad de miniaturización del resonador de cuarzo hace posible la caracterización de múltiples sensores integrados en una configuración en array, esto permitirá en un futuro alcanzar el reto de la detección multianalito para aplicaciones High-Throughput Screening (HTS). / [CA] Durant les dues últimes dècades s'han proposat diferents interfases electrònics per a mesurar els paràmetres més importants de caracterització dels cristalls de microbalança de quars (QCM). La mesura dels paràmetres adequats del sensor per a una aplicació específica és molt important, perquè un error en la interpretació dels resultats pot resultar en un error en la interpretació dels resultats. Els requeriments del sistema de caracterització depenen de l'aplicació. En aquesta tesi, es proposen dos sistemes de caracterització per a dos àmbits d'aplicació que comprenen la majoria de les aplicacions amb sensors QCM: 1) Caracterització de materials sota condicions d'amortiment variable i 2) detecció de substàncies amb alta resolució de mesura. Els sistemes proposats tracten de resoldre la problemàtica detectada en els ja existents. Per a aplicacions en les quals l'amortiment del sensor varia durant l'experiment, es proposa un sistema basat en una nova configuració de la tècnica de compensació automàtica de capacitat (ACC). La nova configuració proporciona la mesura de la freqüència de ressonància sèrie, la resistència dinàmica i la capacitat paral¿lel del sensor. A més, permet un calibratge fàcil del sistema que millora la precisió de la mesura. Es presenten els resultats experimentals per a cristalls de 9 i 10 MHz en mitjans fluids, amb diferents capacitats en paral¿lel, demostrant l'efectivitat de la compensació de capacitat. El sistema presenta alguna desviació en freqüència respecte a la freqüència ressonància sèrie, mesurada amb un analitzador d'impedàncies. Aquestes desviacions són explicades convenientment, degudes al comportament no ideal específic d'alguns components del circuit. Una nova proposta de circuit es presenta com a possible solució a aquest problema. Per a aplicacions d'alta resolució es proposa una plataforma integrada per a caracteritzar sensors acústics d'alta freqüència. El sistema proposat es basa en un nou concepte en el qual el sensor és interrogat mitjançant una font externa molt estable i de molt baix soroll, a una freqüència constant mentre es monitoritzen els canvis produïts per la càrrega en la fase del sensor. L'ús de sensors d'alta freqüència augmenta la sensibilitat de la mesura, per altra banda, el sistema de caracterització dissenyat redueix el soroll en la mateixa. El resultat és una millora en el límit de detecció (LOD). S'aconsegueix amb això un dels reptes pendents en els dispositius acústics d'alta freqüència. La validació de la plataforma desenvolupada es realitza amb una aplicació d'un immunosensor basat en cristalls QCM d'alta freqüència fonamental (HFF-QCM) per a la detecció de dos pesticides: carbaryl i tiabendazol. Els resultats obtinguts per al carbaryl es comparen amb els obtinguts amb altra tecnologia acústica d'alta freqüència basada en sensors Love, amb la tècnica òptica basada en ressonància superficial de plasmons (SPR) i amb la tècnica de referència Enzyme Linked Immuno Assay (ELISA). El LOD obtingut amb els sensors acústics HFF-QCM i Love és similar al obtingut amb les tècniques ELISA i millora en un ordre de magnitud el obtingut amb SPR. La senzillesa conceptual del sistema proposat junt amb el seu baix cost, així com la capacitat de miniaturització del ressonador de quars fa possible la caracterització de múltiples sensors integrats en una configuració en array, el que permetrà en un futur assolir el repte de la detecció multianalit per a aplicacions High-Throughput Screening (HTS). / García Narbón, JV. (2016). Improved characterization systems for quartz crystal microbalance sensors: parallel capacitance compensation for variable damping conditions and integrated platform for high frequency sensors in high resolution applications [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/63249 / Compendio

The lowest natural frequencies of thin-walled noncircular fiber-reinforced composite cylinders, specifically cylinders with elliptical cross sections, are investigated. Of interest is the variation of the lowest natural frequency, the so-called fundamental frequency, as a function of wall laminate properties, cross-sectional eccentricity and other cylinder geometric parameters. Both simple and clamped support boundary conditions are investigated. Laminate properties that are uniform with circumferential location and laminate properties that vary with circumferential location, by way of varying laminate fiber angle with circumferential location, are considered. As the radius of curvature of a noncircular cylinder varies with circumferential location, it is logical to consider the influence of circumferentially varying fiber orientation on the fundamental frequency. The analysis for predicting the fundamental frequency is based on Donnell shell theory, linear elastic properties, and the use of Hamilton's Principle in conjunction with the Rayleigh-Ritz technique. By use of a so-called shape factor, the magnitude of cylinder normal displacements are modulated to be larger in the regions of the cross section with the largest radius of curvatures and smaller in the regions with the smallest radius of curvature. The final equations for predicting the fundamental frequency are quite complex, but a series of approximations results in a hierarchy of simpler equations, the simplest being referred to as Lo's approximation. The prediction of the fundamental frequencies is spot checked by comparing the results as predicted by the various levels of approximation with predictions of a shell-based finite element model. Considering uniform laminate properties, comparisons between the developed analysis and the finite element model are good for all levels of simpler equations, and excellent in some cases. The developed analysis is subsequently used for parameter studies. It is found that compared to a circular cylinder of the same circumference and with uniform laminate properties, the fundamental frequency of an elliptical cylinder is always less. Surprisingly, based on the results obtained, it appears that for a given cylinder geometry the fundamental frequency is not particularly sensitive to wall lamination sequence, though the wave number in the circumferential direction of the mode shape associated with the fundamental frequency is sensitive to lamination sequence. Considering cylinders with circumferentially varying fiber orientation, comparisons between the developed analysis and the finite element model for most of the cases studied are good. However, the developed equations are limited since it is difficult to find a set of known functions to describe the deformation of an arbitrary lamination sequence when applying the Rayleigh-Ritz technique. In general, in can be concluded that the effect of varying fiber orientation on the fundamental frequency is much less than the influence of cylinder aspect ratio. It can also be concluded that the developed analysis would be an excellent tool for design purposes, as the calculation of the fundamental frequency is done quickly, and design trade-offs studies would be easy. / Ph. D.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-05-22T12:39:01Z No. of bitstreams: 1 patriciaribeirodovallecoutinho.pdf: 6874504 bytes, checksum: dcfac5bfe827dd30517321e4731c09ce (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-05-22T17:30:12Z (GMT) No. of bitstreams: 1 patriciaribeirodovallecoutinho.pdf: 6874504 bytes, checksum: dcfac5bfe827dd30517321e4731c09ce (MD5) / Made available in DSpace on 2017-05-22T17:30:12Z (GMT). No. of bitstreams: 1 patriciaribeirodovallecoutinho.pdf: 6874504 bytes, checksum: dcfac5bfe827dd30517321e4731c09ce (MD5) Previous issue date: 2010-03-15 / Prosódia e cognição são os temas centrais do presente trabalho. Reivindica-se o lugar da fonologia na representação formal de construções gramaticais, uma vez que emparelhadas com informações semânticas, sintáticas e pragmáticas estão as informações fonológicas. A partir de contribuições teóricas da Linguística Cognitiva, o conceito de fonema é revisto, sendo tratado em termos de categoria de base corpórea e unidade construcional. Para essa nova visão, é fundamental, ao tratar do fenômeno, uma concepção que envolva a aquisição, o uso e a frequência de construções da língua. Ao se direcionar a atenção para o nível suprassegmental, enfatiza-se a função que a prosódia desempenha na interpretação e na negociação de sentidos. Defende-se que há uma ação conjunta e coordenada na produção da melodia da fala. A fim de analisar como a entoação pode apontar para a construção do significado, foi realizada, através de um programa computacional (PRAAT), a leitura acústica de dados de fala semiespontânea. Os resultados da submissão dos dados foram descritos no sentido de se observar as curvas da Frequência Fundamental (F0) nos trechos que encerram as respostas de entrevistados em coletivas de imprensa. A hipótese é a de que os contornos melódicos nessa transição de turno são descendentes e uma das razões para isso é o desgaste da pressão subglotal que ocorre no fluxo da fala. Outro ponto discutido é que a ação das pregas vocais pode ser considerada um caso particular entre as gestualidades produzidas pelo nosso aparato físico e perceptual. Estabelecendo uma projeção metafórica entre o domínio da coletiva de imprensa e o domínio da guerra, aposta-se na analogia entre os gestos produzidos pelas pregas vocais e os gestos dos braços em um combate. Enfim, quanto mais gestos "sonoros", de alta F0, ocorrem, maior é o poder de combate interacional do falante, na luta pelo turno. Se a F0 decai, dá-se a entrega do turno. São movimentos de caráter físico que estão estreitamente relacionados com o simbolismo conceptual da linguagem. / Prosody and cognition are the central subjects of the present work. The place of phonology in the formal representation of constructions is demanded, since phonological information is paired with semantic, syntactic and pragmatic information. From theoretical contributions of Cognitive Linguistic, the concept of phoneme is reviewed and reformulated with the notions of phoneme as an embodied category and phoneme as a constructional unit. For this new vision, it is fundamental a conception that involves the acquisition, the use and the frequency of language. To direct the attention for the suprassegmental level, it is emphasized the function that the prosody performs in the interpretation and management of meaning. One defends that there is a joint and coordinated action in the production of speech s melody. In order to analyze how the intonation can point out to the construction of meaning, an acoustics reading of data was accomplished through a computation program (PRAAT). The results of the submission of the data were described in order to observe the curves of fundamental frequency (F0) in the stretches that close the interviewed answers in the press conference. The work hypothesis is that the turn transition is configured by falling contours and one of the reasons for this is the deterioration of the subglotal pressure in the flow of the speech. Another point is that the action of vocal folds can be considered a particular case between the gestures produced by our physical and perceptual apparatus. From a metaphorical extension of the press conference's domain and the war's domain, it can be speculated the analogy between the gestures produced by the focal folds and the gestures of the arms in a combat.