BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//LORIA - ECPv6.18.4.1//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://www.loria.fr
X-WR-CALDESC:Évènements pour LORIA
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20250330T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20251026T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20260329T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20261025T010000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:20270328T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:20271031T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20261125T093000
DTEND;TZID=Europe/Paris:20261125T113000
DTSTAMP:20261119T090131Z
CREATED:20261119T090131Z
LAST-MODIFIED:20261119T090131Z
UID:30114-1795599000-1795606200@www.loria.fr
SUMMARY:Soutenance de thèse : Raphaël Bagat (Multispeech)
DESCRIPTION:Raphaël Bagat\, doctorant dans l’équipe Multispeech\, soutiendra sa thèse le 25 novembre à 9 heures 30 en salle C005. \nSa présentation sera en anglais. \nSa thèse est intitulée « Domain Adaptation for Air Traffic Control Automatic Speech Recognition with Limited Transcribed Data« .\n\nDefense committee:\n\nReviewer: Yannick Estève\, Professor\, Avignon Université\, France\nReviewer: Benjamin Lecouteux\, Professor\, Université Grenoble Alpes\, France\nExaminer: Marie Tahon\, Professor\, Le Mans Université\, France\nExaminer: Dietrich Klakow\, Professor\, Université de la Sarre\, Germany\nSupervisor: Irina Illina\, Associate Professor\, Université de Lorraine\, France\nCo-supervisor: Emmanuel Vincent\, Senior Research Scientist\, Centre Inria de l’Université de Lorraine\, France\nGuest: Junichi Yamagishi\, Professor\, National Institute of Informatics\, Japan\nGuest: Raphaël Duroselle\, Ph.D.\, Inria Défense et Sécurité\, France\n\n\nAbstract:\nRecent large-scale pre-trained automatic speech recognition (ASR) systems achieve strong performance across a wide range of conditions\, yet their accuracy can substantially degrade when deployed in specialized domains that differ from their training data\, such as the Air Traffic Control (ATC) domain. This thesis investigates domain adaptation strategies for air traffic control ASR\, where transcribed data are limited. \nOne of the important limitations in ASR adaptation is the high cost of human supervision. Transcribing speech requires domain expertise. ATC communications present acoustic and linguistic variability\, including non-native accents\, diverse operational roles\, noise\, and bandwidth-limited radio channels. Collecting sufficient labeled data to cover all these factors is impractical. \nSince different non-native accents exhibit distinct characteristics\, pooling them into a single adaptation may fail to capture accent-specific patterns\, while maintaining a separate model for each accent does not scale. We propose Mixture of Accent-Specific LoRAs\, which combines accent-specific low-rank experts within a shared model. Our results show that this specialization improves multi-accent speech recognition and remains beneficial even when the speaker’s accent is unknown at inference. \nWhile self-supervised learning with untranscribed data is commonly used to pre-train speech encoders before supervised training\, exploiting it to adapt an already supervised encoder-decoder model\, such as Whisper\, is less straightforward because modifying the encoder may disrupt its compatibility with the pre-trained decoder. To address this issue\, we propose BEST-RQ Encoder Adaptation with Re-training and Distillation\, an adaptation method combining a self-supervised objective with knowledge distillation. BEARD adapts Whisper’s encoder while preserving its compatibility with the pre-trained decoder. BEARD significantly improves supervised adaptation. \nThe third contribution exploits generative data augmentation. Since the generation of additional speech data is only beneficial when the synthetic data faithfully capture the characteristics of the target domain\, it is essential to account for both speaker variability and acoustic conditions. We propose a generative augmentation framework combining text-to-speech\, voice conversion\, accent conversion and an ATC acoustic simulation. Moreover\, we introduce a novel controllable L1-to-L2 accent conversion system. We show that synthetic speech provides the greatest benefits when it reproduces accent diversity and the acoustic characteristics of the target radio channel. \nLinguistic knowledge can be exploited when facing acoustic ambiguity. External language models provide a way of introducing an additional linguistic prior during decoding. We show that external LLMs improve recognition on general-domain speech through reranking and shallow fusion. Then\, we study their use for ATC speech recognition. \nSince ATC communications provide contextual cues such as callsigns and preceding dialogue\, this information can help recognize the current utterance. We leverage Whisper’s native prompting mechanism to provide domain-specific information at inference. We show that prompting can successfully exploit contextual information such as callsigns and dialogue history. \nFinally\, we combine two complementary adaptation strategies. BEARD exploits untranscribed acoustic information during adaptation while prompting exploits contextual information at inference. Their combination allows us to investigate whether these two sources of information are complementary. By combining self-supervised encoder adaptation with contextual prompting\, we achieve the best recognition performance on ATC data reported in this thesis.
URL:https://www.loria.fr/event/soutenance-de-these-raphael-bagat-multispeech/
LOCATION:C005
CATEGORIES:Soutenance
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20261125T093000
DTEND;TZID=Europe/Paris:20261125T123000
DTSTAMP:20261120T150731Z
CREATED:20261120T150731Z
LAST-MODIFIED:20261120T150731Z
UID:30128-1795599000-1795609800@www.loria.fr
SUMMARY:Soutenance de thèse : Raphaël Bagat (équipe Multispeech)
DESCRIPTION:Raphaël Bagat\, doctorant dans l’équipe Multispeech\, soutiendra sa thèse le 25 novembre 2026 à 9h30\, au Loria\, en salle C005\, avec un exposé intitulé  \n« Domain Adaptation for Air Traffic Control Automatic Speech Recognition with Limited Transcribed Data « 
URL:https://www.loria.fr/event/soutenance-de-these-raphael-bagat-multispeech-2/
LOCATION:C005
CATEGORIES:Soutenance
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Paris:20261130T090000
DTEND;TZID=Europe/Paris:20261130T110000
DTSTAMP:20261119T085905Z
CREATED:20261119T085855Z
LAST-MODIFIED:20261119T085905Z
UID:30111-1796029200-1796036400@www.loria.fr
SUMMARY:Soutenance de thèse : Lisa Formentini (Loreley)
DESCRIPTION:Lisa Formentini\, doctorante dans l’équipe Loreley\, soutiendra sa thèse le 30 novembre à 9 heures en salle C005. \nSa présentation sera en anglais. \nLisa Formentini\, PhD student in the Loreley team\, will defend her thesis on 30 November at 9 am in room C005. \n\nHer thesis is entitled “Transformations of Collaborative Ecologies of Artifacts (EoA) in Civil Security Populations”\n\n\nAbstract :\nOrganizational transformation is often viewed through the lens of introducing\, adopting\, or replacing specific technologies. However\, collaborative work rarely relies on isolated artifacts. It is based on evolving configurations of interdependent technologies\, practices\, actors\, and organizational mechanisms. In the field of Computer-Supported Cooperative Work (CSCW)\, the mechanisms by which changes in individual tools and practices extend to the rest of the sociotechnical environment\, as well as the topic of how these environments change over time\, remain relatively understudied. This thesis investigates how collaborative sociotechnical environments are studied and transformed\, and how this understanding can support real-world transformations. Drawing on the concept of Artifact Ecologies (EoA)\, it examines interorganizational collaboration in civil protection through a qualitative field study conducted with a French Departmental Fire and Rescue Service (SDIS) and its partners. The research combines semi-structured interviews and observations of training exercises and field operations\, analyzed thematically\, to conceptualize « movements » and present the first draft of a movement-sensitive framework for digital transformation.\n\nComposition of the jury:\n\n\nThesis supervisors: Pr. Francois CHAROY (Full Professor\, University of Lorraine\, INRIA\, CNRS\, LORIA)\, Dr. Matthieu TIXIER (Associate Professor\, Troyes University of Technology\, LIST3N)\nReviewers: Pr. Catherine LETONDAL (Assistant Professor\, ENAC)\, Pr. Thomas LUDWIG (Full Professor\, University of Siegen)\nExaminers : Dr. Caroline RIZZA (Associate Professor\, Télécom Paris)\, Pr. Marine MINIER (Full Professor\, LORIA\, University of Lorraine)\nGuests: Dr. Samuel NOWAKOWSKI (Maître de conférences\, LORIA\, Université de Lorraine)\, Colonel Dr. Etienne RUDOLF (SDIS57)\n\nKey-words : Ecologies of artifacts\, collaboration\, civil security\, SDIS\, movements
URL:https://www.loria.fr/event/soutenance-de-these-lisa-formentini-loreley/
LOCATION:C005
CATEGORIES:Soutenance
END:VEVENT
END:VCALENDAR