La recerca presentada en aquesta dissertació glossa sobre transformacions de tempo de gravacions monofòniques de saxo jazz preservant l'expressivitat musical. Es una contribució al processament d'audio basat en el contingut, un camp de recerca que ha emergit recentment com a resposta a la necessitat creixent de gestionar intel·ligentment la creixent quantitat d'informació digital multimedia disponible actualment. S'ha investigat com una execució musical, tocada a un tempo concret, es pot reproduir automàticament a un altre tempo mantenint l'expressivitat. Aquest problema no es pot reduir a aplicar una transformació uniforme a totes les notes de la melodia, operació que degradaria la qualitat de l'execució. Proposem un sistema de raonament basat en casos per a transformacions de tempo preservant l'expressivitat. La validació del sistema mostra un comportament superior a la transformació uniforme. A m'es, s'han fet contribucions a l'anàlisi de gravacions expressives, CBR, recuperació de melodies i metodologires d'evaluació de models d'expressivitat. / The research presented in this dissertation focuses on expressivity-aware tempo transformations of monophonic audio recordings of saxophone jazz performances. It is a contribution to content-based audio processing, a field of technology that has recently emerged as an answer to the increased need to deal intelligently with the evergrowing amount of digital multimedia information available nowadays. We have investigated the problem of how a musical performance played at a particular tempo can be rendered automatically at another tempo, while preserving naturally sounding expressivity. This problem cannot be reduced to just applying a uniform transformation to all notes of the melody, since it often degrades the musical quality of the performance. We present a case-based reasoning system for expressivity aware tempo transformations. A validation of the system showed superior results compared to uniform transformation. Furthermore, contributions have been made to expressive performance analysis, CBR, melody retrieval, and evaluation methodologies of expressive models.

Les concepts d'action et de ressource sont omniprésents en informatique. La caractéristique principale d'une action est de changer l'état actuel du système modélisé. Une action peut ainsi être l'exécution d'une instruction dans un programme, l'apprentissage d'un fait nouveau, l'acte concret d'un agent autonome, l'énoncé d'un mot ou encore une tâche planifiée. La caractéristique principale d'une ressource est de pouvoir être divisée, par exemple pour être partagée. Il peut s'agir des cases de la mémoire d'un ordinateur, d'un ensemble d'agents, des différent sens d'une expression, d'intervalles de temps ou de droits d'accès. Actions et ressources correspondent souvent aux dimensions temporelles et spatiales du système modélisé. C'est le cas par exemple de l'exécution d'une instruction sur une case de la mémoire ou d'un groupe d'agents qui coopèrent. Dans ces cas, il est possible de modéliser les actions parallèles comme étant des actions opérant sur des parties disjointes des ressources disponibles. Les logiques modales permettent de modéliser les concepts d'action et de ressource. La sémantique relationnelle d'une modalité unaire est une relation binaire permettant d'accéder à un nouvel état depuis l'état courant. Ainsi une modalité unaire correspond à une action. De même, la sémantique d'une modalité binaire est une relation ternaire permettant d'accéder à deux états. En considérant ces deux états comme des sous-états de l'état courant, une modalité binaire modélise la séparation de ressources. Dans cette thèse, nous étudions des logiques modales utilisées pour raisonner sur les actions, les ressources et la concurrence. Précisément, nous analysons la décidabilité et la complexité du problème de satisfaisabilité de ces logiques. Ces problèmes consistent à savoir si une formule donnée peut être vraie. Pour obtenir ces résultats de décidabilité et de complexité, nous proposons des procédures de décision. Ainsi, nous étudions les logiques modales avec des modalités binaires, utilisées notamment pour raisonner sur les ressources. Nous nous intéressons particulièrement à l'associativité. Alors qu'il est généralement souhaitable que la modalité binaire soit associative, puisque la séparation de ressources l'est, cette propriété rend la plupart des logiques indécidables. Nous proposons de contraindre la valuation des variables propositionnelles afin d'obtenir des logiques décidables ayant une modalité binaire associative. Mais la majeure partie de cette thèse est consacrée à des variantes de la logique dynamique propositionnelle (PDL). Cette logiques possède une infinité de modalités unaires structurée par des opérateurs comme la composition séquentielle, l'itération et le choix non déterministe. Nous étudions tout d'abord des variantes de PDL comparables aux logiques temporelle avec branchement. Nous montrons que les problèmes de satisfaisabilité de ces variantes ont la même complexité que ceux des logiques temporelles correspondantes. Nous étudions ensuite en détails des variantes de PDL ayant un opérateur de composition parallèle de programmes inspiré des logiques de ressources. Cet opérateur permet d'exprimer la séparation de ressources et une notion intéressante d'actions parallèle est obtenue par la combinaison des notions d'actions et de séparation. En particulier, il est possible de décrire dans ces logiques des situations de coopération dans lesquelles une action ne peut être exécutée que simultanément avec une autre. Enfin, la contribution principale de cette thèse est de montrer que, dans certains cas intéressants en pratique, le problème de satisfaisabilité de ces logiques a la même complexité que PDL. / The concepts of action and resource are ubiquitous in computer science. The main characteristic of an action is to change the current state of the modeled system. An action may be the execution of an instruction in a program, the learning of a new fact, a concrete act of an autonomous agent, a spoken word or a planned task. The main characteristic of resources is to be divisible, for instance in order to be shared. Resources may be memory cells in a computer, performing agents, different meanings of a phrase, time intervals or access rights. Together, actions and resources often constitute the temporal and spatial dimensions of a modeled system. Consider for instance the instructions of a computer executed at memory cells or a set of cooperating agents. We observe that in these cases, an interesting modeling of concurrency arises from the combination of actions and resources: concurrent actions are actions performed simultaneously on disjoint parts of the available resources. Modal logics have been successful in modeling both concepts of actions and resources. The relational semantics of a unary modality is a binary relation which allows to access another state from the current state. Hence, unary modalities are convenient to model actions. Similarly, the relational semantics of a binary modality is a ternary relation which allows to access two states from the current state. By interpreting these two states as substates of the current state, binary modalities allow to divide states. Hence, binary modalities are convenient to model resources. In this thesis, we study modal logics used to reason about actions, resources and concurrency. Specifically, we analyze the decidability and complexity of the satisfiability problem of these logics. These problems consist in deciding whether a given formula can be true in any model. We provide decision procedures to prove the decidability and state the complexity of these problems. Namely, we study modal logics with a binary modality used to reason about resources. We are particularly interested in the associativity property of the binary modality. This property is desirable since the separation of resources is usually associative too. But the associativity of a binary modality generally makes the logic undecidable. We propose in this thesis to constrain the valuation of propositional variables to make modal logics with an associative binary modality decidable. The main part of the thesis is devoted to the study of variants of the Propositional Dynamic Logic (PDL). These logics features an infinite set of unary modalities representing actions, structured by some operators like sequential composition, iteration and non-deterministic choice. We first study branching time variants of PDL and prove that the satisfiability problems of these logics have the same complexity as the corresponding branching-time temporal logics. Then we thoroughly study extensions of PDL with an operator for parallel composition of actions called separating parallel composition and based on the semantics of binary modalities. This operator allows to reason about resources, in addition to actions. Moreover, the combination of actions and resources provides a convenient expression of concurrency. In particular, these logics can express situations of cooperation where some actions can be executed only in parallel with some other actions. Finally, our main contribution is to prove that the complexity of the satisfiability problem of a practically useful variant of PDL with separating parallel composition is the same as the satisfiability problem of plain PDL.

La présente thèse est consacrée à l’étude de la catégorie des noms du genre commun dans deux langues slaves : le russe et l'ukrainien. Les noms de cette catégorie ont deux particularités. Premièrement, la plupart d’entre eux sont expressifs et font partie des axiologiques. Nous montrons que leur expressivité peut dépendre de leur contenu phonétique et morphologique ainsi que des constructions syntaxiques dans lesquelles ces noms sont employés. Deuxièmement, les noms de cette catégorie présentent des particularités morphosyntaxiques : ils peuvent appeler l’accord syntaxique ainsi que l’accord sémantique. Nous montrons que le choix du genre de l’accord non seulement satisfait au besoin de référence, mais a également une fonction pragmatique. / SummaryThe present work deals with the category of common gender in two Slavic languages: Russian and Ukrainian. The nouns of this category have two major peculiarities. Firstly, most of them are expressive and axiological nouns. We reveal that the expressivity of this nouns depends on their phonetical and morphological content and on the type of syntactical constructions in which this nouns occur. Secondly, the nouns of this category present morphosyntactical peculiarities: the agreement with this nouns can be semantical and grammatical. We reveal that the choice of the gender of agreement depends not only on semantic and reference, it has also a pragmatic function.

The central focus of this Master's thesis is to describe the ways the elements of linguistic expressivity used in political speeches are transformed in French-to-Czech simultaneous interpreting. In the introductory part of the present thesis, we provide a summary of previous research on the means of expressivity in political speeches, which is completed by a description of the process of simultaneous interpreting. In this way, we intend to define possibilities of transfer of expressive elements in the French-to-Czech language combination. The second part consists of an in-depth contrastive analysis of original French and Czech political speeches followed by a comparison of the former with their interpreted versions. In the third step, we compared the original Czech speeches with interpretations. Using our own methodology, we assessed collected data and came to the conclusion that in French-to- Czech simultaneous interpreting, expressivity rate decreases. Powered by TCPDF (www.tcpdf.org)

Dissertation deal about integration poetry with locomotion in kinderkarten. I investigate measure of interest the children about poetry and integration poetry with locomotion. I suppose, the children will be prefer the integration of activities because of congenital needed and komplex perceive. I engage charakteristic of poetry and locomotion, relation the children to poetry and importance poetry and locomotion for children. I implement the exploration in Prague 1 with twenty children. They are five years old. The method of exploration is observation and experiment. The exploration remain ten days. I investigace measure of interest about poetry and integration poetry with locomotion. I discovered, that I connect poetry and locomotion, the measure of interest increase to sewenteen percent. Thanks to, the children love poetry and get to poetry deap emotional relation. Integration poetry and locomotion attend to harmonious development of personality. I think, that it ensure the best start for life.

The thesis project concerns the development and evaluation of a real-time music visualization system aimed at creating a multi-modal perceptual experience of music emotions. The purpose of the project is to provide singers with real-time visual feedback on their singing, to enhance their expression of emotions in the music. Built upon results from previous studies on emotional expression in music, crossmodal correspondences, and associations among sound, shape, color, and emotions, a singing voice visualization system is proposed that generates real-time graphics to reflect the emotional expression in the input singing in an intuitive fashion. A mapping between musical and visual features was established and tested within a user study regarding the setting of its polarities. The singing voice visualization system was developed as as a software system that runs on personal computers, utilizing Pure Data and Unity. This implementation allows for instantaneous feedback to the user during their singing. The mapping was evaluated in a user study, where participants engage in expressive singing to test the system,in order to assess the meaningfulness of the visual feedback and the effectiveness of the mapping, as well as the impact of the polarity. The results show that color as a strong visual cue of emotional expression provided meaningful information on some participants’ expression of typical happiness and sadness. Other cues of the visual feedback possibly enhanced some participants’ emotional expression of singing in an indirect way. The polarity had a noticeable impact on the perception of the visual feedback. The current study is limited by the reliability of the techniques used for the extracting acoustic features from real-time singing, particularly in the detection of attack speed. The evaluation were limited by the broad definition of one of the research questions. The findings of this study suggest potential applications for the singing voice visualization system in fields of music education, art, and entertainment. Additionally, the research highlights the need for further exploration and refinement in the design of the mapping and the evaluation methodology. / Avhandlingsprojektet handlar om utveckling och utvärdering av ett realtidssystem för musikvisualisering som syftar till att skapa en multimodal perceptuell upplevelse av musikkänslor. Syftet med projektet är att ge sångare visuell feedback i realtid på deras sång, för att förstärka deras uttryck av känslor i musiken. Med utgångspunkt i resultat från tidigare studier om känslouttryck i musik, korsmodala korrespondenser och associationer mellan ljud, form, färg och känslor, föreslås ett visualiseringssystem för sångstämmor som genererar realtidsgrafik för att på ett intuitivt sätt återspegla känslouttrycket i den inmatade sången. En mappning mellan musikaliska och visuella funktioner etablerades och testades i en användarstudie avseende inställningen av dess polariteter. Visualiseringssystemet för sångröst utvecklades som ett mjukvarusystem som körs på persondatorer med Pure Data och Unity. Denna implementering möjliggör omedelbar feedback till användaren under dennes sång. Kartläggningen utvärderades i en användarstudie, där deltagarna fick sjunga uttrycksfullt för att testa systemet, för att bedöma meningsfullheten i den visuella feedbacken och kartläggningens effektivitet, samt polaritetens inverkan. Resultaten visar att färg som en stark visuell signal för känslomässiga uttryck gav meningsfull information om vissa deltagares uttryck av typisk lycka och sorg. Andra ledtrådar i den visuella återkopplingen förstärkte möjligen vissa deltagares känslomässiga uttryck av sång på ett indirekt sätt. Polariteten hade en märkbar inverkan på uppfattningen av den visuella feedbacken. Den aktuella studien begränsas av tillförlitligheten hos de tekniker som används för att extrahera akustiska egenskaper från sång i realtid, särskilt när det gäller att upptäcka attackhastighet. Utvärderingen begränsades av den breda definitionen av en av forskningsfrågorna. Resultaten av denna studie visar på potentiella tillämpningar för visualiseringssystemet för sångröster inom musikutbildning, konst och underhållning. Dessutom belyser forskningen behovet av ytterligare utforskning och förfining av kartläggningens utformning och utvärderingsmetodiken.

Deep neural networks (DNN) have been widely demonstrated to be more powerful than their shallower counterparts in a variety of computer vision tasks and remote sensing applications. However, as many techniques are based on trial-and-error experiments as opposed to systematic evaluation, scientific evidence for the superiority of DNN needs more theoretical and experimental foundations. Recent work has shown that the neural network expressivity, measured by the number of linear regions, is independent of the network structure, suggesting that the success of deep neural networks is attributed to its ease of training. Inspired by this, this project aims to investigate novel approaches to train neural networks and obtain desired properties of the regional properties of linear regions. In particular, it highlights the regional structure of linear regions in different decision regions and seeks to initialize the network in a better position that makes it easier to have this regional structure. By counting the total number of linear regions in the input space, we validated that the shallow wider networks and the deep narrow networks share the same upper-bound expressivity in different synthetic datasets. We also discovered that the linear regions along the decision boundary are larger in shape and fewer in number, while being denser and fitted to the data manifold when close to the data. Our experiments indicate that the proposed initialization method can generate more linear regions at initialization, make the training converge faster, and finally generate linear regions that better fit the data manifold on synthetic data. On the EuroSAT satellite dataset, the proposed initialization method does not facilitate the convergence of ResNet-18, but achieves better performance with an average increase of 0.14% on accuracy compared to pre-trained weights and 0.19% compared to He uniform initialization. / Djupa neurala nätverk (Deep Neural Networks, DNN) har i stor utsträckning visat sig vara mer kraftfulla än sina grunda motsvarigheter i en mängd olika datorseendeuppgifter och fjärranalystillämpningar. Många tekniker är dock baserade på försök och misstag snarare än systematisk utvärdering, och vetenskapliga bevis för DNN:s överlägsenhet behöver mer teoretiska och experimentella grunder. Nyligen utförda arbeten har visat att det neurala nätverkets uttrycksförmåga, mätt som antalet linjära regioner, är oberoende av nätverksstrukturen, vilket tyder på att framgången för djupa neurala nätverk beror på att de är lätta att träna. Inspirerat av detta syftar detta projekt till att undersöka nya metoder för att träna neurala nätverk och få önskade egenskaper hos de regionala egenskaperna hos linjära regioner. I synnerhet belyser det den regionala strukturen hos linjära regioner i olika beslutsregioner och försöker initiera nätverket i ett bättre läge som gör det lättare att få denna regionala struktur. Genom att räkna det totala antalet linjära regioner i ingångsutrymmet validerade vi att de grunda bredare nätverken och de djupa smala nätverken har samma övre gräns för uttrycklighet i olika syntetiska dataset. Vi upptäckte också att de linjära regionerna längs beslutsgränsen är större till formen och färre till antalet, samtidigt som de är tätare och anpassade till datamångfalden när de ligger nära data. Våra experiment visar att den föreslagna initialiseringsmetoden kan generera fler linjära regioner vid initialiseringen, få träningen att konvergera snabbare och slutligen generera linjära regioner som bättre passar datamångfalden på syntetiska data. På EuroSAT-satellitdatamängden underlättar den föreslagna initialiseringsmetoden inte konvergensen för ResNet-18, men uppnår bättre prestanda med en genomsnittlig ökning av noggrannheten med 0,14% jämfört med förtränade vikter och 0,19% jämfört med He uniform initialisering.

No description available.

L'objectif de la thèse est de proposer des méthodes permettant d'améliorer l'expressivité des systèmes de synthèse de la parole. Une des propositions centrales de ce travail est de définir, utiliser et mesurer l'impact de structures linguistiques opérant au delà du niveau de la phrase, par opposition aux approches opérant sur des phrases isolées de leur contexte. Le cadre de l'étude est restreint au cas de la lecture de contes pour enfants. Les contes ont la particularité d'avoir été l'objet d'un certain nombre d'études visant à en dégager une structure narrative et de faire intervenir une certain nombre de stéréotypes de personnages (héros, méchant, fée) dont le discours est souvent rapporté. Ces caractéristiques particulières sont exploitées pour modéliser les propriétés prosodiques des contes au delà du niveau de la phrase. La transmission orale des contes a souvent été associée à une pratique musicale (chants, instruments) et leur lecture reste associée à des propriétés mélodiques très riches, dont la reproduction reste un défi pour les synthétiseurs de parole modernes. Pour répondre à ces problématiques, un premier corpus de contes écrits est collecté et annoté avec des informations relatives à la structure narrative des contes, l'identification et l'attribution des citations directes, le référencement des mentions des personnages ainsi que des entités nommées et des énumérations étendues. Le corpus analysé est décrit en terme de couverture et d'accord inter-annotateurs. Il est utilisé pour modéliser des systèmes de segmentation des contes en épisode, de détection des citations directes, des actes de dialogue et des modes de communication. Un deuxième corpus de contes lus par un locuteur professionnel est présenté. La parole est alignée avec les transcriptions lexicale et phonétique, les annotations du corpus texte et des méta-informations décrivant les caractéristiques des personnages intervenant dans le conte. Les relations entre les annotations linguistiques et les propriétés prosodiques observées dans le corpus de parole sont décrites et modélisées. Finalement, un prototype de contrôle des paramètres expressifs du synthétiseur par sélection d'unités Acapela est réalisé. Le prototype génère des instructions prosodiques opérant au delà du niveau de la phrase, notamment en utilisant les informations liées à la structure du conte et à la distinction entre discours direct et discours rapporté. La validation du prototype de contrôle est réalisée dans le cadre d'une expérience perceptive, qui montre une amélioration significative de la qualité de la synthèse. / The aim of this thesis is to propose ways to improve the expressiveness of speech synthesis systems. One of the central propositions of this work is to define, use and measure the impact of linguistic structures operating beyond the sentence level, as opposed to approaches operating on sentences out of their context. The scope of the study is restricted to the case of storytelling for children. The stories have the distinction of having been the subject of a number of studies in order to highlight a narrative structure and involve a number of stereotypical characters (hero, villain, fairy) whose speech is often reported. These special features are used to model the prosodic properties tales beyond the sentence level. The oral transmission of tales was often associated with musical practice (vocals, instruments) and their reading is associated with rich melodic properties including reproduction remains a challenge for modern speech synthesizers. To address these issues, a first corpus of written tales is collected and annotated with information about the narrative structure of stories, identification and allocation of direct quotations, referencing references to characters as well as named entities and enumerations areas. The corpus analyzed is described in terms of coverage and inter-annotator agreement. It is used to model systems segmentation tales episode, detection of direct quotes, dialogue acts and modes of communication. A second corpus of stories read by a professional speaker is presented. The word is aligned with the lexical and phonetic transcriptions, annotations of the corpus text and meta-information describing the characteristics of the characters involved in the story. The relationship between linguistic annotations and prosodic properties observed in the speech corpus are described and modeled. Finally, a prototype control expressive synthesizer parameters by Acapela unit selection is made. The prototype generates prosodic operating instructions beyond the sentence level, including using the information related to the structure of the story and the distinction between direct speech and reported speech. Prototype validation control is performed through a perceptual experience, which shows a significant improvement in the quality of the synthesis.

Esta dissertação tem dois objetivos básicos: primeiro, investigar se há marcas de oralidade na linguagem narrativa e das personagens do livro Quarto de despejo: diário de uma favelada, de Carolina Maria de Jesus (1914-1977); segundo, examinar como essas manifestações se processam linguisticamente no texto no plano lexical. Suscitam-se, portanto, duas perguntas: Primeira: Há em Quarto de despejo manifestações linguísticas da oralidade? Segunda: Em caso afirmativo, em que níveis essa oralidade se processa linguisticamente? Para responder a essas perguntas, temos a hipótese de que a língua falada foi aproveitada de maneira mais ou menos inconsciente pela autora de Quarto de despejo e que, entre vários níveis linguísticos, a oralidade se manifesta de maneira notável em nível lexical, por meio de um vocabulário de uso popular. Estudar-se-á como o léxico da linguagem narrativa e das personagens apresenta efeitos de língua falada, tornando o texto dinâmico e expressivo. Este trabalho está teoricamente fundamentado em pressupostos da Sociolinguística e da Estilística. A metodologia de trabalho envolveu várias leituras da obra, anotações de vocabulário e análises à luz da teoria proposta. Estabelecem-se relações entre fala e escrita, compreendidas não como uma dicotomia, mas como polos de um continuum tipológico. Nesse continuum, um gênero da escrita pode apresentar características da fala e vice-versa. Ressalte-se que, tendo em vista esse referencial teórico, a narrativa de Quarto de despejo não é considerada um texto oral, mas um texto escrito em cuja enunciação se evidencia a presença da oralidade graças a um leque de vocábulos expressivos. / This thesis has two basic goals: first, to investigate whether there are marks of orality in the narrative language and lines from the characters from the book \"Child of the dark: The diary of Carolina Maria de Jesus\", by Carolina Maria de Jesus (1914-1977); secondly, to examine how these manifestations are processed linguistically into the text on the lexical sphere. Two questions, therefore, are raised: First: are there linguistic manifestations of orality in Child of the dark? Second: if so, at what levels is this orality processed linguistically? To answer those questions, we hypothesized that the spoken language was used more or less unconsciously by the author of Child of the dark, and that between several linguistic levels, orality manifests itself in a remarkable manner in the lexical level, by use of a colloquial vocabulary. We shall study how the lexicon of the narrative language and of the characters has effects of the spoken language, making the text dynamic and expressive. This paper is theoretically grounded in assumptions of Sociolinguistic and Stylistic. The work methodology involved several readings of the book, vocabulary annotations and analysis in the light of the proposed theory. Relationships between speech and writing are set, understood not as a dichotomy but as poles of a typological continuum. In this continuum, a genre of writing can display characteristics of speech and vice versa. It is emphasized that, in view of that theoretical framework, the narrative of Child of the dark is not regarded as an oral text, but as a written text where enunciation highlights the presence of orality using a variety of expressive vocabularies.