Search

BACKGROUND Treatment for patients with recurrent ovarian cancer has been mainly based on systemic therapy. The role of secondary cytoreductive surgery is unclear. METHODS We randomly assigned patients with recurrent ovarian cancer who had a first relapse after a platinum-free interval (an interval during which no platinum-based chemotherapy was used) of 6 months or more to undergo secondary cytoreductive surgery and then receive platinum-based chemotherapy or to receive platinumbased chemotherapy alone. Patients were eligible if they presented with a positive Arbeitsgemeinschaft Gynäkologische Onkologie (AGO) score, defined as an Eastern Cooperative Oncology Group performance-status score of 0 (on a 5-point scale, with higher scores indicating greater disability), ascites of less than 500 ml, and complete resection at initial surgery. A positive AGO score is used to identify patients in whom a complete resection might be achieved. The primary end point was overall survival. We also assessed quality of life and prognostic factors for survival. RESULTS A total of 407 patients underwent randomization: 206 were assigned to cytoreductive surgery and chemotherapy, and 201 to chemotherapy alone. A complete resection was achieved in 75.5% of the patients in the surgery group who underwent the procedure. The median overall survival was 53.7 months in the surgery group and 46.0 months in the no-surgery group (hazard ratio for death, 0.75; 95% confidence interval, 0.59 to 0.96; P = 0.02). Patients with a complete resection had the most favorable outcome, with a median overall survival of 61.9 months. A benefit from surgery was seen in all analyses in subgroups according to prognostic factors. Quality-of-life measures through 1 year of follow-up did not differ between the two groups, and we observed no perioperative mortality within 30 days after surgery. CONCLUSIONS In women with recurrent ovarian cancer, cytoreductive surgery followed by chemotherapy resulted in longer overa

In the context of the optimization of Deep Neural Networks, we propose to rescale the learning rate using a new technique of automatic differentiation. This technique relies on the computation of the {\em curvature}, a second order information whose computational complexity is in between the computation of the gradient and the one of the Hessian-vector product. If (1C,1M) represents respectively the computational time and memory footprint of the gradient method, the new technique increase the overall cost to either (1.5C,2M) or (2C,1M). This rescaling has the appealing characteristic of having a natural interpretation, it allows the practitioner to choose between exploration of the parameters set and convergence of the algorithm. The rescaling is adaptive, it depends on the data and on the direction of descent. The numerical experiments highlight the different exploration/convergence regimes.

Depuis les années 80, les codecs vocaux reposent sur des stratégies de codage à court terme qui fonctionnent au niveau de la sous-trame ou de la trame (généralement 5 à 20 ms). Les chercheurs ont essentiellement ajusté et combiné un nombre limité de technologies disponibles (transformation, prédiction linéaire, quantification) et de stratégies (suivi de forme d'onde, mise en forme du bruit) pour construire des architectures de codage de plus en plus complexes. Dans cette thèse, plutôt que de s'appuyer sur des stratégies de codage à court terme, nous développons un cadre alternatif pour la compression de la parole en codant les attributs de la parole qui sont des caractéristiques perceptuellement importantes des signaux vocaux. Afin d'atteindre cet objectif, nous résolvons trois problèmes de complexité croissante, à savoir la classification, la prédiction et l'apprentissage des représentations. La classification est un élément courant dans les conceptions de codecs modernes. Dans un premier temps, nous concevons un classifieur pour identifier les émotions, qui sont parmi les attributs à long terme les plus complexes de la parole. Dans une deuxième étape, nous concevons un prédicteur d'échantillon de parole, qui est un autre élément commun dans les conceptions de codecs modernes, pour mettre en évidence les avantages du traitement du signal de parole à long terme et non linéaire. Ensuite, nous explorons les variables latentes, un espace de représentations de la parole, pour coder les attributs de la parole à court et à long terme. Enfin, nous proposons un réseau décodeur pour synthétiser les signaux de parole à partir de ces représentations, ce qui constitue notre dernière étape vers la construction d'une méthode complète de compression de la parole basée sur l'apprentissage automatique de bout en bout. Bien que chaque étape de développement proposée dans cette thèse puisse faire partie d'un codec à elle seule, chaque étape fournit également des informations et une ba, Since the 80s, speech codecs have relied on short-term coding strategies that operate at the subframe or frame level (typically 5 to 20ms). Researchers essentially adjusted and combined a limited number of available technologies (transform, linear prediction, quantization) and strategies (waveform matching, noise shaping) to build increasingly complex coding architectures. In this thesis, rather than relying on short-term coding strategies, we develop an alternative framework for speech compression by encoding speech attributes that are perceptually important characteristics of speech signals. In order to achieve this objective, we solve three problems of increasing complexity, namely classification, prediction and representation learning. Classification is a common element in modern codec designs. In a first step, we design a classifier to identify emotions, which are among the most complex long-term speech attributes. In a second step, we design a speech sample predictor, which is another common element in modern codec designs, to highlight the benefits of long-term and non-linear speech signal processing. Then, we explore latent variables, a space of speech representations, to encode both short-term and long-term speech attributes. Lastly, we propose a decoder network to synthesize speech signals from these representations, which constitutes our final step towards building a complete, end-to-end machine-learning based speech compression method. The first two steps, classification and prediction, provide new tools that could replace and improve elements of existing codecs. In the first step, we use a combination of source-filter model and liquid state machine (LSM), to demonstrate that features related to emotions can be easily extracted and classified using a simple classifier. In the second step, a single end-to-end network using long short-term memory (LSTM) is shown to produce speech frames with high subjective quality for packet loss concealment (PL

This paper presents a new neural speech compression method that is practical in the sense that it operates at low bitrate, introduces a low latency, is compatible in computational complexity with current mobile devices, and provides a subjective quality that is comparable to that of standard mobile-telephony codecs. Other recently proposed neural vocoders also have the ability to operate at low bitrate. However, they do not produce the same level of subjective quality as standard codecs. On the other hand, standard codecs rely on objective and short-term metrics such as the segmental signal-to-noise ratio that correlate only weakly with perception. Furthermore, standard codecs are less efficient than unsupervised neural networks at capturing speech attributes, especially long-term ones. The proposed method combines a cognitive-coding encoder that extracts an interpretable unsupervised hierarchical representation with a multi stage decoder that has a GAN-based architecture. We observe that this method is very robust to the quantization of representation features. An AB test was conducted on a subset of the Harvard sentences that are commonly used to evaluate standard mobile-telephony codecs. The results show that the proposed method outperforms the standard AMR-WB codec in terms of delay, bitrate and subjective quality.

In this paper we introduce CorticalFlow, a new geometric deep-learning model that, given a 3-dimensional image, learns to deform a reference template towards a targeted object. To conserve the template mesh's topological properties, we train our model over a set of diffeomorphic transformations. This new implementation of a flow Ordinary Differential Equation (ODE) framework benefits from a small GPU memory footprint, allowing the generation of surfaces with several hundred thousand vertices. To reduce topological errors introduced by its discrete resolution, we derive numeric conditions which improve the manifoldness of the predicted triangle mesh. To exhibit the utility of CorticalFlow, we demonstrate its performance for the challenging task of brain cortical surface reconstruction. In contrast to current state-of-the-art, CorticalFlow produces superior surfaces while reducing the computation time from nine and a half minutes to one second. More significantly, CorticalFlow enforces the generation of anatomically plausible surfaces; the absence of which has been a major impediment restricting the clinical relevance of such surface reconstruction methods.

In this paper we introduce CorticalFlow, a new geometric deep-learning model that, given a 3-dimensional image, learns to deform a reference template towards a targeted object. To conserve the template mesh’s topological properties, we train our model over a set of diffeomorphic transformations. This new implementation of a flow Ordinary Differential Equation (ODE) framework benefits from a small GPU memory footprint, allowing the generation of surfaces with several hundred thousand vertices. To reduce topological errors introduced by its discrete resolution, we derive numeric conditions which improve the manifoldness of the predicted triangle mesh. To exhibit the utility of CorticalFlow, we demonstrate its performance for the challenging task of brain cortical surface reconstruction. In contrast to current state-of-the-art, CorticalFlow produces superior surfaces while reducing the computation time from nine and a half minutes to one second. More significantly, CorticalFlow enforces the generation of anatomically plausible surfaces; the absence of which has been a major impediment restricting the clinical relevance of such surface reconstruction methods.

Dans le domaine du codage audio, l’utilisation d’une extension de bande paramétrique pour encoder les hautes fréquences permet une économie de débit. Contrairement à la majorité des méthodes d'extension de bande qui effectuent leur traitement dans le domaine temporel, le traitement proposé dans ce mémoire s’effectue dans le domaine fréquentiel pour créer une extension de bande audio de 8 à 16 kHz sur les signaux monophoniques de parole et de musique. Le projet de maîtrise étudie plusieurs éléments : les techniques de régénération des hautes fréquences, l'utilisation du cepstre pour le calcul et la représentation de l'enveloppe spectrale, ainsi que des techniques d’amélioration de la dynamique du spectre. Les résultats de tests subjectifs formels montrent un gain sur certaines catégories de signaux par rapport au codec AMR-WB+, lequel est reconnu comme étant un standard international développé à l'Université de Sherbrooke.

Conventional feature-based classification methods do not apply well to automatic recognition of speech emotions, mostly because the precise set of spectral and prosodic features that is required to identify the emotional state of a speaker has not been determined yet. This paper presents a method that operates directly on the speech signal, thus avoiding the problematic step of feature extraction. Furthermore, this method combines the strengths of the classical source-filter model of human speech production with those of the recently introduced liquid state machine (LSM), a biologically-inspired spiking neural network (SNN). The source and vocal tract components of the speech signal are first separated and converted into perceptually relevant spectral representations. These representations are then processed separately by two reservoirs of neurons. The output of each reservoir is reduced in dimensionality and fed to a final classifier. This method is shown to provide very good classification performance on the Berlin Database of Emotional Speech (Emo-DB). This seems a very promising framework for solving efficiently many other problems in speech processing.

We propose an approach for cognitive coding of speech by unsupervised extraction of contextual representations in two hierarchical levels of abstraction. Speech attributes such as phoneme identity that last one hundred milliseconds or less are captured in the lower level of abstraction, while speech attributes such as speaker identity and emotion that persist up to one second are captured in the higher level of abstraction. This decomposition is achieved by a two-stage neural network, with a lower and an upper stage operating at different time scales. Both stages are trained to predict the content of the signal in their respective latent spaces. A top-down pathway between stages further improves the predictive capability of the network. With an application in speech compression in mind, we investigate the effect of dimensionality reduction and low bitrate quantization on the extracted representations. The performance measured on the LibriSpeech and EmoV-DB datasets reaches, and for some speech attributes even exceeds, that of state-of-the-art approaches.

Dans le domaine du codage audio, l’utilisation d’une extension de bande paramétrique pour encoder les hautes fréquences permet une économie de débit. Contrairement à la majorité des méthodes d'extension de bande qui effectuent leur traitement dans le domaine temporel, le traitement proposé dans ce mémoire s’effectue dans le domaine fréquentiel pour créer une extension de bande audio de 8 à 16 kHz sur les signaux monophoniques de parole et de musique. Le projet de maîtrise étudie plusieurs éléments : les techniques de régénération des hautes fréquences, l'utilisation du cepstre pour le calcul et la représentation de l'enveloppe spectrale, ainsi que des techniques d’amélioration de la dynamique du spectre. Les résultats de tests subjectifs formels montrent un gain sur certaines catégories de signaux par rapport au codec AMR-WB+, lequel est reconnu comme étant un standard international développé à l'Université de Sherbrooke.

L’écoféminisme repose sur la prémisse selon laquelle l’oppression des femmes et la domination de la nature sont issues des mêmes structures sociétales sous-jacentes. L’écoféminisme tisse donc un lien entre enjeux féministes et crise climatique. L’intégration de la perspective de genre dans la lutte et l’adaptation aux changements climatiques est nécessaire tout comme l’ajout de perspectives écoféministes dans l’élaboration de politiques sensibles au genre qui incluent les principes de l'égalité climatique. L'urgence climatique nécessite un changement radical de la part des gouvernements, basé sur la reconnaissance du rôle crucial des femmes dans la recherche et la mise en œuvre de solutions, et ce, malgré le fait qu’elles soient parmi les plus vulnérables face aux changements climatiques et à la dégradation environnementale.

We propose a novel learning based algorithm to generate efficient and physically plausible sampling patterns in MRI. This method has a few advantages compared to recent learning based approaches: i) it works off-the-grid and ii) allows to handle arbitrary physical constraints. These two features allow for much more versatility in the sampling patterns that can take advantage of all the degrees of freedom offered by an MRI scanner. The method consists in a high dimensional optimization of a cost function defined implicitly by an algorithm. We propose various numerical tools to address this numerical challenge.

In this paper we compute the Hausdorff distance between sets of continuous curves and sets of piecewise constant or linear discretizations. These sets are Sobolev balls given by the continuous or discrete $L^p$-norm of the derivatives. We detail the suitable discretization or smoothing procedure which are preservative in the sense of these norms. Finally we exhibit the link between Eulerian numbers and the uniformly space knots B-spline used for smoothing.

L’écoféminisme jette la lumière sur le lien étroit entre le féminisme et l’écologie. Ce courant de pensée repose sur le postulat selon lequel l’oppression des femmes et la destruction de la nature sont intrinsèquement reliées en ce qu’elles trouvent racine dans les mêmes structures de domination, patriarcales et capitalistes. Bien qu’à l’origine les contours de ce courant furent tracés par Françoise d’Eaubonne au début des années 1970, l’écoféminisme s’est véritablement construit autour de deux évènements historiques principaux, soit la révolution verte de l’agriculture en Inde et la lutte antinucléaire aux États-Unis. Les périodes entourant ces deux évènements sont abordées dans le cadre du présent essai, au sein d’une mise en contexte visant à retracer l’émergence de l’écoféminisme. L’objectif poursuivi consiste à démontrer les apports de l’écoféminisme dans une perspective d’adaptation aux changements climatiques, tout en portant une attention particulière au contexte autochtone canadien. En matière d’adaptation aux changements climatiques, la reconnaissance du lien existant entre genre et climat est fondamentale. Ce lien est d’autant plus éloquent dans les pays dits « du Sud », où les femmes sont directement dépendantes des ressources naturelles afin d’assurer la subsistance de leur famille. La question de l’accès à l’eau en est un exemple éloquent. Au cours des dernières décennies, plusieurs mouvements menés par des femmes en faveur de la protection de l’environnement ont montré que les femmes doivent être considérées, non pas uniquement comme étant particulièrement vulnérables aux effets des changements climatiques, mais également comme se trouvant en ligne de front de la lutte environnementale. En outre, certains évènements récents relatifs à la politique canadienne permettent de tisser des liens entre les enjeux écoféministes et autochtones, notamment en matière d’extractivisme et plus particulièrement le cas de l’expansion du pipeline Trans Mountain. En pol

Carbonates form ubiquitously throughout the history of deposition, burial, and uplift of basins. As such, they potentially record the environmental conditions at the time of formation. Carbonate clumped isotopes provide the temperature of precipitation but can be internally reordered if the host rock is exposed to elevated temperatures over geologic time scales. Here, we exploited this kinetic behavior by analyzing multiple generations of cements that capture the range of environments experienced by the El Abra Formation from eastern Mexico. From this, we developed a quantitative diagenetic history for these different phases of cementation. We observed a 70 °C range in clumped isotope temperatures from 64 °C to 134 °C for these cements, which is not compatible with their inferred precipitation environments. This suggests that bond reordering occurred during burial but did not fully reorder all cements to a common apparent temperature. We reconstructed original cement growth temperatures and the isotopic signature of the parent fluids to show that precipitation from a marine pore fluid began at 125 Ma, contemporaneous with deposition, and continued throughout burial to temperatures of at least 138 °C at 42 Ma. We show that precipitation of equant cements, which occluded 90% of the pore space, was coincident with Laramide-related burial to depths greater than 3800 m. A U-Pb age of diagenetic calcite of 77.1 ± 3.6 Ma provides independent support for our estimates of the absolute timing of precipitation of two distinct phases of the paragenesis. This is the first demonstration of the utility of integrating U-Pb age dating with reordered clumped isotope temperatures to provide quantitative constraints on the time-temperature history of cementation. Such information may ultimately lead to advances in our understanding of the formational environments and geological processes that drive diagenesis in carbonates for temperatures below the clumped isotope "blocking temperatur

Carbonates form ubiquitously throughout the history of deposition, burial, and uplift of basins. As such, they potentially record the environmental conditions at the time of formation. Carbonate clumped isotopes provide the temperature of precipitation but can be internally reordered if the host rock is exposed to elevated temperatures over geologic time scales. Here, we exploited this kinetic behavior by analyzing multiple generations of cements that capture the range of environments experienced by the El Abra Formation from eastern Mexico. From this, we developed a quantitative diagenetic history for these different phases of cementation. We observed a 70 °C range in clumped isotope temperatures from 64 °C to 134 °C for these cements, which is not compatible with their inferred precipitation environments. This suggests that bond reordering occurred during burial but did not fully reorder all cements to a common apparent temperature. We reconstructed original cement growth temperatures and the isotopic signature of the parent fluids to show that precipitation from a marine pore fluid began at 125 Ma, contemporaneous with deposition, and continued throughout burial to temperatures of at least 138 °C at 42 Ma. We show that precipitation of equant cements, which occluded 90% of the pore space, was coincident with Laramide-related burial to depths greater than 3800 m. A U-Pb age of diagenetic calcite of 77.1 ± 3.6 Ma provides independent support for our estimates of the absolute timing of precipitation of two distinct phases of the paragenesis. This is the first demonstration of the utility of integrating U-Pb age dating with reordered clumped isotope temperatures to provide quantitative constraints on the time-temperature history of cementation. Such information may ultimately lead to advances in our understanding of the formational environments and geological processes that drive diagenesis in carbonates for temperatures below the clumped isotope "blocking temperatur

The Runge approximation theorem for holomorphic maps is a fundamental result in complex analysis, and, consequently, many works have been devoted to extend it to other spaces (e.g. maps between certain algebraic varieties or complex manifolds). This article presents such a result for pseudo-holomorphic maps from a compact Riemann surface to a compact almost-complex manifold M, given that the manifold M admits many pseudo-holomorphic maps from $${\mathbb {C}{\rm P}^1}$$ which can be thought of as local approximations of the Laurent expansion az +br 2/z. This result specializes to some compact algebraic varieties (e.g. rationally connected projective varieties). An application to Lefschetz fibrations is presented

We establish a result which states that regularizing an inverse problem with the gauge of a convex set $C$ yields solutions which are linear combinations of a few extreme points or elements of the extreme rays of $C$. These can be understood as the \textit{atoms} of the regularizer. We then explicit that general principle by using a few popular applications. In particular, we relate it to the common wisdom that total gradient variation minimization favors the reconstruction of piecewise constant images., Comment: in Proceedings of iTWIST'18, Paper-ID: 30, Marseille, France, November, 21-23, 2018

We establish a general principle which states that regularizing an inverse problem with a convex function yields solutions which are convex combinations of a small number of atoms. These atoms are identified with the extreme points and elements of the extreme rays of the regularizer level sets. An extension to a broader class of quasi-convex regularizers is also discussed. As a side result, we characterize the minimizers of the total gradient variation, which was still an unresolved problem.

En traitement du signal audio, les plosives sont des sons de parole très importants au regard de l’intelligibilité et de la qualité. Les plosives sont cependant difficiles à modéliser à l’aide des techniques usuelles (prédiction linéaire et codage par transformée), à cause de leur dynamique propre importante et à cause de leur nature non prédictible. Cette étude présente un exemple de système complet capable de détecter, segmenter, et altérer les plosives dans un flux de parole. Ce système est utilisé afin de vérifier la validité de l’hypothèse suivante : La phase d’éclatement (de burst) des plosives peut être mise à zéro, de façon perceptuellement équivalente. L’impact sur la qualité subjective de cette transformation est évalué sur une banque de phrases enregistrées. Les résultats de cette altération hautement destructive des signaux tendent à montrer que l’impact perceptuel est mineur. Les implications de ces résultats pour le codage de la parole sont abordées.

En traitement du signal audio, les plosives sont des sons de parole très importants au regard de l’intelligibilité et de la qualité. Les plosives sont cependant difficiles à modéliser à l’aide des techniques usuelles (prédiction linéaire et codage par transformée), à cause de leur dynamique propre importante et à cause de leur nature non prédictible. Cette étude présente un exemple de système complet capable de détecter, segmenter, et altérer les plosives dans un flux de parole. Ce système est utilisé afin de vérifier la validité de l’hypothèse suivante : La phase d’éclatement (de burst) des plosives peut être mise à zéro, de façon perceptuellement équivalente. L’impact sur la qualité subjective de cette transformation est évalué sur une banque de phrases enregistrées. Les résultats de cette altération hautement destructive des signaux tendent à montrer que l’impact perceptuel est mineur. Les implications de ces résultats pour le codage de la parole sont abordées.

This chapter looks at leadership and management of multidisciplinary teams in the mental health context. It provides an overview of what constitutes a multidisciplinary team and how policy can change the roles and relationships in teams. The potential conflict inherent in teamwork is outlined. A brief overview of leadership styles is provided, with a more detailed description of the relationship between different leadership styles and team effectiveness and satisfaction in the mental health context. Finally, suggestions about effective leadership styles and practical tips for team building and managing team meetings are provided.

This chapter looks at leadership and management of multidisciplinary teams in the mental health context. It provides an overview of what constitutes a multidisciplinary team and how policy can change the roles and relationships in teams. The potential conflict inherent in teamwork is outlined. A brief overview of leadership styles is provided, with a more detailed description of the relationship between different leadership styles and team effectiveness and satisfaction in the mental health context. Finally, suggestions about effective leadership styles and practical tips for team building and managing team meetings are provided.

This chapter looks at leadership and management of multidisciplinary teams in the mental health context. It provides an overview of what constitutes a multidisciplinary team and how policy can change the roles and relationships in teams. The potential conflict inherent in teamwork is outlined. A brief overview of leadership styles is provided, with a more detailed description of the relationship between different leadership styles and team effectiveness and satisfaction in the mental health context. Finally, suggestions about effective leadership styles and practical tips for team building and managing team meetings are provided.

This chapter looks at leadership and management of multidisciplinary teams in the mental health context. It provides an overview of what constitutes a multidisciplinary team and how policy can change the roles and relationships in teams. The potential conflict inherent in teamwork is outlined. A brief overview of leadership styles is provided, with a more detailed description of the relationship between different leadership styles and team effectiveness and satisfaction in the mental health context. Finally, suggestions about effective leadership styles and practical tips for team building and managing team meetings are provided.

This chapter looks at leadership and management of multidisciplinary teams in the mental health context. It provides an overview of what constitutes a multidisciplinary team and how policy can change the roles and relationships in teams. The potential conflict inherent in teamwork is outlined. A brief overview of leadership styles is provided, with a more detailed description of the relationship between different leadership styles and team effectiveness and satisfaction in the mental health context. Finally, suggestions about effective leadership styles and practical tips for team building and managing team meetings are provided.

Le domaine de recherche de la modification de durée d'un signal audio est actif depuis plus d'une quarantaine d'années. Aujourd'hui, plusieurs applications en font usage: livres audio, synthèse par échantillonnage, voix sur IP, postsynchronisation audio-vidéo, etc. Dans la plupart des cas, l'objectif des techniques de modification de durée est de modifier le débit du signal tout en y affectant le moins possible ses caractéristiques perceptuelles (tel que la tonalité). Plusieurs algorithmes de modification de durée ont été proposés pour tenter d'atteindre ce but. Ceux-ci sont caractérisés par des forces et faiblesses selon le type de signal traité. Dans le cas où l'on tente de traiter un signal de type variable, l'ensemble des techniques proposées à ce jour exhibent des artéfacts audibles qui affectent l'ensemble de la qualité subjective du signal d'origine. Une étude approfondie de l'état de l'art permet de constater que les algorithmes de modification de durée peuvent être classés sous deux familles, soit les techniques appliquées dans le domaine temporel et les techniques appliquées dans le domaine fréquentiel. Selon le domaine de traitement choisi (temporel ou fréquentiel), la qualité subjective obtenue prime sur un type de signal particulier. Par exemple, les techniques appliquées dans le domaine temporel ont la caractéristique d'être efficaces sur des signaux monophoniques, tandis que les techniques appliquées dans le domaine fréquentiel sont plus performantes sur des signaux polyphoniques ou bruités. Ce constat permet d'établir qu'il existe une complémentarité entre les techniques appliquées dans le domaine temporel et le domaine fréquentiel. Ceci motive la création d'un algorithme de modification de durée qui en tire avantage, de façon à résoudre le problème du traitement de signaux composés de plusieurs types. Cet ouvrage présente une méthode novatrice qui vise à exploiter la complémentarité observée entre les techniques appliquées dans le domaine temporel e

Le domaine de recherche de la modification de durée d'un signal audio est actif depuis plus d'une quarantaine d'années. Aujourd'hui, plusieurs applications en font usage: livres audio, synthèse par échantillonnage, voix sur IP, postsynchronisation audio-vidéo, etc. Dans la plupart des cas, l'objectif des techniques de modification de durée est de modifier le débit du signal tout en y affectant le moins possible ses caractéristiques perceptuelles (tel que la tonalité). Plusieurs algorithmes de modification de durée ont été proposés pour tenter d'atteindre ce but. Ceux-ci sont caractérisés par des forces et faiblesses selon le type de signal traité. Dans le cas où l'on tente de traiter un signal de type variable, l'ensemble des techniques proposées à ce jour exhibent des artéfacts audibles qui affectent l'ensemble de la qualité subjective du signal d'origine. Une étude approfondie de l'état de l'art permet de constater que les algorithmes de modification de durée peuvent être classés sous deux familles, soit les techniques appliquées dans le domaine temporel et les techniques appliquées dans le domaine fréquentiel. Selon le domaine de traitement choisi (temporel ou fréquentiel), la qualité subjective obtenue prime sur un type de signal particulier. Par exemple, les techniques appliquées dans le domaine temporel ont la caractéristique d'être efficaces sur des signaux monophoniques, tandis que les techniques appliquées dans le domaine fréquentiel sont plus performantes sur des signaux polyphoniques ou bruités. Ce constat permet d'établir qu'il existe une complémentarité entre les techniques appliquées dans le domaine temporel et le domaine fréquentiel. Ceci motive la création d'un algorithme de modification de durée qui en tire avantage, de façon à résoudre le problème du traitement de signaux composés de plusieurs types. Cet ouvrage présente une méthode novatrice qui vise à exploiter la complémentarité observée entre les techniques appliquées dans le domaine temporel e

Voice communication systems are moving at a steady pace from circuit-switched to packet-switched networks. Converged networks, in which data traffic and multimedia traffic share the same infrastructures, have yet to face many challenges. Unlike data traffic, voice traffic is very sensitive to issues such as delay, jitter and packet losses. The Quality of Service (QoS) of IP networks has greatly improved over the recent years. Nevertheless, it is not uncommon to experience loss in QoS on some networks caused by lack of resources or software/hardware failures. The speech codec (COder DECoder) is an essential component in a digital voice communication system. It converts the speech signal into a compact representation in order to efficiently use the available bandwidth. To compress the signal, speech codecs usually exploit the high redundancy present in a speech signal using techniques known as predictive coding. The most widely used speech coding model is the Code Excited Linear Prediction (CELP) model upon which are based most of the modern standardized speech codecs. The use of predictive codebooks, the Adaptive CodeBook (ACB) in particular, allows a tremendous reduction in bitrate. Unfortunately, the extensive use of prediction renders CELP codecs more vulnerable to packet losses than non-predictive codecs. A possible solution is to avoid using prediction at the cost of an increased bitrate. In this thesis, we propose several techniques to reduce the vulnerability of CELP codecs to packet losses caused by the use of prediction. Solutions that aim at limiting the contribution of the ACB and thus that reduce the sensitivity of the codec to packet losses are presented and discussed. Techniques based on sending side information, in order to help the decoder recover after a packet loss, are also proposed. The solutions are constrained to be low delay and low bitrate solutions, and are required to maintain interoperability if used in a standardized codec. Objective and s

The ANTARES neutrino telescope is a large photomultiplier array designed to detect neutrino-induced upward-going muons by their Cherenkov radiation. Understanding the absorption and scattering of light in the deep Mediterranean is fundamental to optimising the design and performance of the detector. This paper presents measurements of blue and UV light transmission at the ANTARES site taken between 1997 and 2000. The derived values for the scattering length and the angular distribution of particulate scattering were found to be highly correlated, and results are therefore presented in terms of an absorption length;,ab, and an effective scattering length lambda(sct)(eff). The values for blue (UV) light are found to be lambda(abs) similar or equal to 60(26) m, lambda(sct)(eff similar or equal to) 265(122) m, with significant (similar to15%) time variability. Finally, the results of ANTARES simulations showing the effect of these water properties on the anticipated performance of the detector are presented. (C) 2004 Elsevier B.V. All rights reserved.

The aim of this paper is to chart the relevant studies and research questions concerning the relationship between ICTs and Mobility in Everyday Life – a concept defined and delineated in a previous paper An Agenda for ‘Mobility in Everyday Life’ for ICT Researchers (Haddon, 2000a).

The aim of this paper is to chart the relevant studies and research questions concerning the relationship between ICTs and Mobility in Everyday Life – a concept defined and delineated in a previous paper An Agenda for ‘Mobility in Everyday Life’ for ICT Researchers (Haddon, 2000a).

Collection : Perspectives anglo-saxonnes, Collection : Perspectives anglo-saxonnes, Contient une table des matières, Avec mode texte

Contient une table des matières, Avec mode texte, Entretiens

Contient une table des matières, Avec mode texte, Entretiens

Comprend : Version du second livre de l' :Aeneïde: +Éneïde+

Comprend : Version du second livre de l' :Aeneide: +Énéide+, Appartient à l’ensemble documentaire : HNormand1, Appartient à l’ensemble documentaire : Rtmgus1