The explosive growth of Internet-based architectures is allowing an efficient access to information resources over geographically dispersed areas. This fact is exerting a major influence on current manufacturing practices. Business activities involving customers, partners, employees and suppliers are being rapidly and efficiently integrated through networked information management environments. Therefore, efforts are required to take advantage of distributed infrastructures that can satisfy information integration and collaborative work strategies in corporate environments. In this research, Internet-based distributed solutions focused on the manufacturing industry are proposed. Three different systems have been developed for the tooling sector, specifically for the company Seco Tools UK Ltd (industrial collaborator). They are summarised as follows. SELTOOL is a Web-based open tool selection system involving the analysis of technical criteria to establish appropriate selection of inserts, toolholders and cutting data for turning, threading and grooving operations. It has been oriented to world-wide Seco customers. SELTOOL provides an interactive and crossed-way of searching for tooling parameters, rather than conventional representation schemes provided by catalogues. Mechanisms were developed to filter, convert and migrate data from different formats to the database (SQL-based) used by SELTOOL.TTS (Tool Trials System) is a Web-based system developed by the author and two other researchers to support Seco sales engineers and technical staff, who would perform tooling trials in geographically dispersed machining centres and benefit from sharing data and results generated by these tests. Through TTS tooling engineers (authorised users) can submit and retrieve highly specific technical tooling data for both milling and turning operations. Moreover, it is possible for tooling engineers to avoid the execution of new tool trials knowing the results of trials carried out in physically distant places, when another engineer had previously executed these trials. The system incorporates encrypted security features suitable for restricted use on the World Wide Web. An urgent need exists for tools to make sense of raw data, extracting useful knowledge from increasingly large collections of data now being constructed and made available from networked information environments. This explosive growth in the availability of information is overwhelming the capabilities of traditional information management systems, to provide efficient ways of detecting anomalies and significant patterns in large sets of data. Inexorably, the tooling industry is generating valuable experimental data. It is a potential and unexplored sector regarding the application of knowledge capturing systems. Hence, to address this issue, a knowledge discovery system called DISKOVER was developed. DISKOVER is an integrated Java-application consisting of five data mining modules, able to be operated through the Internet. Kluster and Q-Fast are two of these modules, entirely developed by the author. Fuzzy-K has been developed by the author in collaboration with another research student in the group at Durham. The final two modules (R-Set and MQG) have been developed by another member of the Durham group. To develop Kluster, a complete clustering methodology was proposed. Kluster is a clustering application able to combine the analysis of quantitative as well as categorical data (conceptual clustering) to establish data classification processes. This module incorporates two original contributions. Specifically, consistent indicators to measure the quality of the final classification and application of optimisation methods to the final groups obtained. Kluster provides the possibility, to users, of introducing case-studies to generate cutting parameters for particular Input requirements. Fuzzy-K is an application having the advantages of hierarchical clustering, while applying fuzzy membership functions to support the generation of similarity measures. The implementation of fuzzy membership functions helped to optimise the grouping of categorical data containing missing or imprecise values. As the tooling database is accessed through the Internet, which is a relatively slow access platform, it was decided to rely on faster Information retrieval mechanisms. Q-fast is an SQL-based exploratory data analysis (EDA) application, Implemented for this purpose.

There are many students in the class that have dyslexia and can struggle with simple tasks such as reading. Therefore, this study set out to investigate the applications of text to speech synthesizers facilitate learning English at upper secondary with these students in focus from a teacher’s perspective. This study has been conducted through means of a semi- structured interviews with secondary school teachers. Research to support the fact that TTS does in fact aid facilitate the reading of students with difficulties has been stark. One the other hand scholars have claimed that it does not always aid struggling readers, therefore this paper will explore the discrepancies between these contrasting views. In the results it was found that the usage of TTS in the classroom should be seen as a compensatory tool that can aid struggling students in reading as opposed to being seen as a solution. The question to whether this can aid students without struggling difficulties was bound to the intelligibility of the voice of the TTS. Furthermore, it was maintained that this could in fact encourage students with their reading on the basis of academic success.

Speech technology can facilitate human-machine interaction and create new communication interfaces. Text-To-Speech (TTS) systems provide speech output for dialogue, notification and reading applications as well as personalized voices for people that have lost the use of their own. TTS systems are built to produce synthetic voices that should sound as natural, expressive and intelligible as possible and if necessary be similar to a particular speaker. Although naturalness is an important requirement, providing the correct information in adverse conditions can be crucial to certain applications. Speech that adapts or reacts to different listening conditions can in turn be more expressive and natural. In this work we focus on enhancing the intelligibility of TTS voices in additive noise. For that we adopt the statistical parametric paradigm for TTS in the shape of a hidden Markov model (HMM-) based speech synthesis system that allows for flexible enhancement strategies. Little is known about which human speech production mechanisms actually increase intelligibility in noise and how the choice of mechanism relates to noise type, so we approached the problem from another perspective: using mathematical models for hearing speech in noise. To find which models are better at predicting intelligibility of TTS in noise we performed listening evaluations to collect subjective intelligibility scores which we then compared to the models’ predictions. In these evaluations we observed that modifications performed on the spectral envelope of speech can increase intelligibility significantly, particularly if the strength of the modification depends on the noise and its level. We used these findings to inform the decision of which of the models to use when automatically modifying the spectral envelope of the speech according to the noise. We devised two methods, both involving cepstral coefficient modifications. The first was applied during extraction while training the acoustic models and the other when generating a voice using pre-trained TTS models. The latter has the advantage of being able to address fluctuating noise. To increase intelligibility of synthetic speech at generation time we proposed a method for Mel cepstral coefficient modification based on the glimpse proportion measure, the most promising of the models of speech intelligibility that we evaluated. An extensive series of listening experiments demonstrated that this method brings significant intelligibility gains to TTS voices while not requiring additional recordings of clear or Lombard speech. To further improve intelligibility we combined our method with noise-independent enhancement approaches based on the acoustics of highly intelligible speech. This combined solution was as effective for stationary noise as for the challenging competing speaker scenario, obtaining up to 4dB of equivalent intensity gain. Finally, we proposed an extension to the speech enhancement paradigm to account for not only energetic masking of signals but also for linguistic confusability of words in sentences. We found that word level confusability, a challenging value to predict, can be used as an additional prior to increase intelligibility even for simple enhancement methods like energy reallocation between words. These findings motivate further research into solutions that can tackle the effect of energetic masking on the auditory system as well as on higher levels of processing.

Text-till-talsynteser (eng. Text-to-speech, TTS), med andra ord digitalt skapade röster, har blivit en naturlig del av våra dagliga liv. Vi hör dem i kollektivtrafiken, när vi interagerar med digitala assistenter och när människor som saknar en egen röst använder dem för att kommunicera. Dessa artificiella röster har, precis som riktiga röster, olika egenskaper som påverkar hur de låter och uppfattas. Det är därför intressant att förstå vilka TTS-synteser som användare föredrar i olika sammanhang. I denna studie undersöktes huruvida det finns en preferens för olika kön hos instruerande TTS-synteser. I studien fick 46 deltagare spela ett spel där målet var att fullfölja ett antal upplästa instruktioner genom olika knapptryck på skärmen. Spelet var uppdelat i två omgångar, där instruktören i ena omgången var en TTS-syntes med en manlig röst och i den andra en kvinnlig. Vilken TTS-syntes som började och ordning av instruktioner slumpades för deltagarna. Efter spelad omgång svarade deltagarna på en enkät om upplevda röstegenskaper hos instruktören. När sista omgången fullföljts besvarades även en avslutande enkät där det bland annat frågades vilken av TTS-synteserna de föredrog, följt av en kvalitativ fråga där de fick motivera sitt svar. Svaren analyserades sedan genom ett statistiskt test och en kvalitativ analys. Studien kunde ej påvisa någon preferens för kön hos instruerande TTS-synteser. Istället visade resultatet att testpersonerna har individuella preferenser och åsikter om rösterna, även om det verkade finnas viss konsensus om somliga upplevda röstegenskaper hos TTS-synteser. / Text-to-speech synthesis (TTS), in other words digitally created voices, has become a natural part of our daily lives. We hear them in public transport, when we interact with digital assistants and when people who do not have their own voice use them to communicate. These artificial voices, just like real voices, have different properties that affect how they sound and are perceived. It is thereby interesting to understand which TTS-synthesis users prefer in different contexts. This study examined whether there is a preference for different genders in instructional TTS-synthesis. In the study, 46 participants played a game where the goal was to complete a number of read out loud instructions through various key presses on the screen. The game was divided into two rounds, where the instructor in one round was a TTS-synthesis with a male voice and female in the other one. Which TTS-synthesis started and the order of instructions was randomized for each participant. Upon completing the round, the participants answered a questionnaire about perceived voice characteristics of the instructor. When the final round was completed, a final questionnaire was also answered in which, among other things, it was asked which of the TTS-syntheses the test subjects preferred, followed by a qualitative question in which they were asked to justify their answer. The answers were then analyzed through a statistical test and a qualitative analysis. The study could not demonstrate any gender preference in instructional TTS-synthesis. Instead, the results showed that the test subjects have individual preferences and opinions about the voices, even though there seemed to be some consensus about some of the perceived voice characteristics of the TTS-syntheses.

No description available.

Lors du fonctionnement d'un turboréacteur, la rotation de l'axe du moteur induit une force centrifuge qui plaque les aubes contre les alvéoles du disque, le long de surfaces appelées « portées ». Au cours de la vie du moteur, des micro-glissements (« fretting ») sont générés au niveau des portées. Ce travail de thèse s'intéresse aux endommagements de type oligocyclique des portées aube/ disque du compresseur Haute Pression (HP) d'un turboréacteur civil. Afin de réduire le frottement et de protéger les pièces, les portées des aubes sont revêtues d'un dépôt épais de CuNiln sur lequel est appliqué un lubrifiant solide. Le système est soumis à des pressions fortes (600MPa) et à des températures élevées (500°C). Des expertises sur pièces ont permis d'identifier des usures importantes et des phénomènes de transfert. Pour simuler expérimentalement le contact, un moyen d'essai de fretting usure en température a été développé dans le cadre de cette thèse. A partir d'une géométrie de contact échelle 1 simplifiée, choisie pour représenter les champs de chargement moteur, le domaine des sollicitations a été balayé et les comportements tribologiques identifiés et caractérisés. L'étude tribologique a permis de mettre en évidence l'influence forte de la force normale sur les mécanismes d'endommagements, dont les scénarios de mise en place ont pu être explicités pour les deux domaines majeurs de sollicitations. Dans un deuxième temps, les endommagements ont été étudiés d'un point de vue physicochimique et le rôle tribochimique joué par les éléments environnementaux (0, N ... ) a été mis en évidence. Pour finir, la comparaison directe des mécanismes identifiés sur banc expérimental et sur le système industriel a permis d'aborder, d'un point de vue critique, l'importance des paramètres choisis pour les équivalences expérience / système industriel ; et de la dynamique de chargement pour garantir la représentativité d'un essai à l'échelle laboratoire. / In the blade/ disk contact of aero-engines, movements induced by the engine regime changes may lead to critical wear. In order to limit friction and protect components, motorists usually use a thick and soft CuNiin layer (150 !-lm) covered with a lubricant. This system is subjected to high contact pressure (upto 600 MPa) and high temperatures (up to 500 °C). In service observations of engine parts revealed significant wear and transfer phenomena. An experimental layout was designed to perform fretting wear tests in temperature on a punch/plane contact, representative of engine solicitation. Based on the tribological study performed, specified gradation mechanisms were identified and the evidence of a pressure threshold was demonstrated. Physicochemical analyses performed confirmed the role played by environment (0, N ... ) in fretting wear mechanisms and suggest that the tribochemical effects should be considered in the fretting wear models. Direct comparisons between degradation mechanisms identified on the experimental system, and on industrial parts, allow a critical review of parameters chosen to simulate the contact. Finally, the impact of loading dynamics was investigated in order to improve tests representativity at la bora tory scale.

Tato práce se zabývá syntézou řeči z textu. V práci je podán základní teoretický úvod do syntézy řeči z textu. Práce je postavena na MARY TTS systému, který umožňuje využít existujících modulů k vytvoření vlastního systému pro syntézu řeči z textu, a syntéze řeči pomocí skrytých Markovových modelů natrénovaných na vytvořené řečové databázi. Bylo vytvořeno několik jednoduchých programů ulehčujících vytvoření databáze a přidání nového jazyka a hlasu pro MARY TTS systém bylo demonstrováno. Byl vytvořen a publikován modul a hlas pro Český jazyk. Byl popsán a implementován algoritmus pro přepis grafémů na fonémy.

Ένα TTS (Τext-To-Speech) σύστημα μετατρέπει ένα οποιοδήποτε κείμενο στην αντιστοιχούσα ομιλία, η οποία έχει φυσικά χαρακτηριστικά. Το ΤΤS αποτελείται από δύο βαθμίδες, τη βαθμίδα Επεξεργασίας Φυσικής Γλώσσας (ΕΦΓ) και τη βαθμίδα Ψηφιακής Επεξεργασίας Σήματος (ΨΕΣ). Η βαθμίδα ΕΦΓ είναι υπεύθυνη για την σωστή ανάλυση του κειμένου εισόδου σε φωνήματα και το καθορισμό των επιθυμητών προσωδιακών χαρακτηριστικών, όπως το pitch, η διάρκεια και η ένταση του κάθε φωνήματος. Η βαθμίδα ΨΕΣ αναλαμβάνει να συνθέσει την ομιλία με τα επιθυμητά προσωδιακά χαρακτηρίστηκα, τα οποία έδωσε η βαθμίδα ΕΦΓ. Ένας τρόπος για να επιτευχθεί αυτό είναι με χρήση αλγορίθμων ανάλυσης και σύνθεσης ομιλίας, όπως ο αλγόριθμος HNM (Harmonic plus Noise Model).Ο ΗΝΜ μοντελοποιεί το σήμα ομιλίας ως άθροισμα δύο τμημάτων, ενός τμήματος με αρμονικά χαρακτηριστικά και ενός τμήματος με χαρακτηριστικά θορύβου. Χρησιμοποιώντας αυτό το μοντέλο γίνεται η ανάλυση και η σύνθεση του σήματος ομιλίας με ή χωρίς προσωδιακές μεταβολές. / A TTS (Text-To-Speech) System is used to convert any given text to its corresponding speech with natural characteristics. A TTS consists of two modules, the Natural Language Processing (NLP) module and the Digital Signal Processing (DSP) module. The NLP module analyses the input text and supplies the DSP module with the appropriate phonemes and prosodic modifications, with concern to pitch, duration and volume of each phoneme. Then the DSP module synthesizes speech with the target prosody, using speech analysis-synthesis algorithms such as HNM. HNM (Harmonic plus Noise Model) algorithm models speech signal as the sum two parts, the harmonic part and the noise part. Speech analysis and speech synthesis with or without modifications, is achieved using the harmonic and the noise part

Dans l’industrie, les traitements mécaniques de surface métalliques permettent d’améliorer les conditions de service des pièces mécaniques. Les effets de contact de ces types de procédés engendrent une forte déformation plastique du matériau et par conséquent une transformation microstructurale en sous-surface. Cette transformation se manifeste dans le raffinement progressif de la microstructure dans une couche de quelques dizaines de micromètres. Celle-ci est souvent dénommé "surface tribologiquement transformée" (en anglais : Tribologically Transformed Surface - TTS). Une telle transformation microstructurale conduit à une augmentation des propriétés mécaniques en extrême surface et rend le matériau plus résistant aux conditions de frottement, usure et fatigue.Dans le cadre de cette étude, deux procédures de transformation microstructurale ont été employées sur un matériau modèle : le fer-α. Pour la première technique (grenaillage), la surface est impactée de façon répétitive avec des billes métalliques projetées à grande vitesse. Concernant la deuxième méthode (micro-percussion), la surface est impactée répétitivement à un endroit précis avec un indenteur conique rigide.L’objet de ce projet se centre sur trois aspects principaux : (i) déterminer les gradients mécaniques et microstructuraux induits sur les deux types de surfaces transformées (grenaillage et micro-percussion), (ii) établir un lien quantitatif entre les mesures faites par deux types d’essais micromécaniques (nano-indentation et micro-compression de piliers) et (iii) mettre en évidence les effets microstructuraux impliqués (taille de grain, densité de dislocations, etc...) dans l’augmentation des propriétés mécaniques par hyper-déformation de surfaces. / The mechanical surface treatments confer better local mechanical properties against wear or fatigue service conditions. In the case of impact-based treatments, the material is exposed to repeated mechanical loadings, producing a severe plastic deformation in the near-surface. It leads to a local and progressive refinement of the microstructure into the affected zone, commonly known as Tribologically Transformed Surface (TTS). For this project, two mechanical surface treatments are used in a model material (pure α-iron): (i) shot-peening and (ii) micro-percussion.The resulting surfaces are characterized by a mechanical property gradient in-depth as a consequence of the microstructural transformation over a few tens of microns. Nowadays, it is well-known that this rise of local mechanical properties could improve the service lifetime of materials. However, a simple micro-hardness test is not quite enough to quantify precisely the engendered variation of mechanical properties and understand the influence of several microstructural effects. For this purpose, two micro-mechanical tests are considered: (i) nano-indentation and (ii) in situ micro-pillar compression.The main issue of this work is to characterize the mechanically-induced transformed surfaces and correlate the mechanical properties gradients with the local microstructural evolutions. Indeed, three main goals are considered: (i) quantify the mechanical and microstructural gradients induced by the surface treatments (shot-peening and micro-percussion), (ii) correlate the results obtained by the means of both mechanical tests (nano-indentation and micro-pillar compression) and finally (iii) investigate the influence of several microstructural effects related with the graded strengthening of hyper-deformed surfaces.

This thesis introduces a general method for incorporating the distributional analysis of textual and linguistic objects into text-to-speech (TTS) conversion systems. Conventional TTS conversion uses intermediate layers of representation to bridge the gap between text and speech. Collecting the annotated data needed to produce these intermediate layers is a far from trivial task, possibly prohibitively so for languages in which no such resources are in existence. Distributional analysis, in contrast, proceeds in an unsupervised manner, and so enables the creation of systems using textual data that are not annotated. The method therefore aids the building of systems for languages in which conventional linguistic resources are scarce, but is not restricted to these languages. The distributional analysis proposed here places the textual objects analysed in a continuous-valued space, rather than specifying a hard categorisation of those objects. This space is then partitioned during the training of acoustic models for synthesis, so that the models generalise over objects' surface forms in a way that is acoustically relevant. The method is applied to three levels of textual analysis: to the characterisation of sub-syllabic units, word units and utterances. Entire systems for three languages (English, Finnish and Romanian) are built with no reliance on manually labelled data or language-specific expertise. Results of a subjective evaluation are presented.