
Orange S.A. (French pronunciation: [ɔʁɑ̃ʒ ɛs‿a]), formerly France Télécom S.A., is a French multinational telecommunications corporation. It has 266 million customers worldwide and employs 89,000 people in France, and 59,000 elsewhere. It is the tenth largest mobile network operator in the world and the fourth largest in Europe after Vodafone, Telefónica and VEON. In 2015, the group had revenue of €40 billion. The company's head office is located in the 15th arrondissement of Paris. The current CEO is Stéphane Richard. The company is a component of the Euro Stoxx 50 stock market index.Orange has been the company's main brand for mobile, landline, internet and IPTV services since 2006. It originated in 1994 when Hutchison Whampoa acquired a controlling stake in Microtel Communications during the early 1990s and rebranded it as "Orange". It became a subsidiary of Mannesmann in 1999 and was acquired by France Télécom in 2000. The company was rebranded as Orange on 1 July 2013.
Spatial audio is spreading in applications such as virtual and augmented reality and immersive games. The higher-order ambisonic (HOA) format is particularly useful in this context. Transmitting spatial information requires multiple channels, e.g., 16 channels for 3rd-order ambisonics, resulting in increased memory requirements for storage and higher bitrates for communication. Therefore, efficient compression algorithms are necessary for those contents. The recently standardized IVAS codec allows the coding of HOA content for communication use-cases. Here, we propose to evaluate it in comparison with a basic multi-mono approach across a variety of contents and spatialization methods. Results show that IVAS outperforms the multi-mono approach at the same bitrate. In particular, this codec exploits inter-channel correlation to reduce the bitrate. We point out that it is therefore especially robust for signals with a high interchannel correlation, such as those composed of a limited number of plane waves. Conversely, the multi-mono approach is unable to exploit this correlation and performs poorly on this type of signal.
The evaluation of voice anonymisation remains challenging. Current practice relies on automatic speaker verification metrics such as the equal error rate (EER). Performance estimates dependent on the classifier and operating point provide an incomplete or even misleading characterisation of privacy risk. We investigate the use of similarity rank disclosure (SRD), an information-theoretic metric, which operates on feature representations rather than classifier decisions, providing a threshold-independent assessment of privacy and analysis of both average and worst-case disclosure. We report its application to speaker embeddings, fundamental frequency, and phone embeddings using 2024 VoicePrivacy Challenge systems. The SRD reveals privacy leaks and system-specific weaknesses missed by EER-based evaluation. Findings highlight the merit of representation-level metrics and demonstrate the potential of SRD as a flexible and interpretable tool for the evaluation of voice anonymisation.
This paper reports an empirical study evaluating the relevance of several RAG metrics. The experiment is based on a question-answering dataset created by human annotators from business data. The generated responses and retrieved spans of a RAG system are scored using evaluation metrics from four libraries (Ragas, DeepEval, RAGChecker, Opik). These metrics are compared to scores given by two evaluators, as well as to standard metrics such as recall. An analysis of correlations is conducted. Finally, we highlight certain limitations of our methodology, compare it to those used in the literature, and suggest some avenues for future research. This paper is an English translation of a paper originally published in the French-speaking workshop EvalLLM (Brabant, 2026).
Les Nouvelles Formes d’Organisation du Travail (NFOT) postulent que le numérique change en totalité la manière de travailler. Cela implique pour les organisations de se repenser entièrement : espaces, outils, management. Ces trois champs sont souvent désignés par l’abréviation « 3B » pour « Bricks - Bytes - Behaviors ». L’idée d’un salarié débarrassé des tâches laborieuses par la technologie illustre l’avenir du travail depuis les années 80. Le déploiement de l’intelligence artificielle générative en fournit la plus récente déclinaison, avec la notion de travailleur « réhumanisé », concentré sur les tâches spécifiquement humaines et « à valeur ajoutée ». Cette mutation peut-elle s’opérer sans effet adverse ? Nous nous appuyons sur une revue de littérature pluridisciplinaire pour identifier les facteurs de ressource et de risque portés par les NFOT.Le récit sur la liberté, la créativité et la productivité qui accompagne le déploiement des NFOT en masque les contraintes potentielles. Ainsi, le flex office peut accroître les stresseurs environnementaux et diminuer les échanges et la coopération. La numérisation porte un risque de dispersion, d’intensification et de moindre autonomie. L’intelligence artificielle fait en particulier s’interroger sur le maintien des compétences. Le Lean Management, le mode projet, l’agilité, dévoyés de leurs principes et environnements initiaux, mènent vers la précarisation subjective, le surengagement et la complexification des relations. Le télétravail peut percuter le collectif. En outre, chacun des « 3B » crée des activités supplémentaires, qui éloignent du travail tout en alourdissant sa charge : trouver le bon espace, choisir le bon canal de communication, valoriser sa production et optimiser ses relations sociales pour maintenir son employabilité interne, acquérir de nouvelles compétences à un rythme toujours plus soutenu.Jusque récemment, il semble que les NFOT n’ont pas stimulé l’innovation et la productivité au niveau attendu. Aujourd’hui, l’IA est présentée comme changeant la donne et des suppressions de postes sont annoncées. Parallèlement, l’incidence des troubles psychosociaux ne cesse de croître. Nous proposons comme piste que les NFOT focalisent sur le « Comment », oubliant les questions : « Quoi ? Pourquoi ? Avec qui ? » et ratent l’étape de la personnalisation en déployant des solutions standardisées. Ainsi, elles peuvent échouer à produire l’optimisation annoncée. Combinant perte de sens, perte d’engagement, perte de culture et de sérendipité, voire perte de compétence, les NFOT peuvent produire de la « DYSERGIE », terme que nous proposons pour en synthétiser les effets socio-organisationnels adverses.
Early Classification of Time Series (ECTS) requires making accurate decisions as early as possible in inherently online and evolving environments. Yet, most existing methods assume stationarity and rely on separable designs, where classification and triggering are optimized independently, an assumption that fundamentally limits their adaptability under drift. In this work, we challenge this paradigm and study ECTS under non-stationary conditions. We provide the first systematic comparison between separable and end-to-end approaches across controlled drifting scenarios. Building on Reinforcement Learning, we introduce DQeND, a unified architecture that jointly learns representation, classification, and triggering decisions, while remaining directly comparable to state-of-the-art separable baselines. Across a wide range of drifts, DQeND demonstrates strong robustness across various non-stationary scenarios, consistently outperforming separable baselines. An ablation study further highlights that jointly updating representation and decision modules is critical to these gains. Overall, our results indicate that end-to-end learning can offer improved adaptation capabilities for ECTS in dynamic environments, and motivate further investigation of alternatives to separable designs.