Théo Mariotte
Présentation
Actuellement Maître de Conférences au Laboratoire d'Informatique de l'Université du Mans (LIUM), mes travaux se concentrent sur le traitement automatique de la parole et du signal audio à l'aide de l'apprentissage profond. Les travaux menés durant ma thèse se sont concentrés sur la segmentation et regroupement en locuteurs (speaker diarization) avec des antennes de microphones. Ensuite, ma recherche s'est orientée vers les modèles d'IA interprétables pour le traitement du signal audio et la séparation de sources dans des conditions acoustiques difficiles et sous contraintes de calcul.
Publications
Publications
|
|
Channel-Combination Algorithms for Robust Distant Voice Activity and Overlapped Speech DetectionIEEE/ACM Transactions on Audio, Speech and Language Processing, In press |
|
|
SPARSE AUTOENCODERS MAKE AUDIO FOUNDATION MODELS MORE EXPLAINABLEInternational Conference on Acoustics, Speech, and Signal Processing, May 2026, Barcelone (Espagne), Spain |
|
|
Multiple Choice Learning for Efficient Speech Separation with Many SpeakersICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Apr 2025, Hyderabad, India. ⟨10.1109/ICASSP49660.2025.10888528⟩ |
|
|
LOCALIZE, FILTER, SEGMENT: TOWARDS MULTI-MICROPHONE SPEAKER SEGMENTATIONForum Acusticum Euronoise 2025, Jun 2025, Malaga, Spain |
|
|
Predefined Prototypes for Intra-Class Separation and DisentanglementInterspeech 2024, ISCA, Sep 2024, Kos / Greece, Greece. pp.3809-3813, ⟨10.21437/Interspeech.2024-825⟩ |
|
|
ASoBO: Attentive Beamformer Selection for Distant Speaker Diarization in MeetingsInterspeech, International Speech Communication Association (ISCA), Sep 2024, Kos / Greece, Greece |
|
|
AN EXPLAINABLE PROXY MODEL FOR MULTILABEL AUDIO SEGMENTATIONInternational Conference on Acoustics Speech and Signal Processing, IEEE, Apr 2024, Seoul (Korea), France |
|
|
Unsupervised multiple domain translation through controlled Disentanglement in variational autoencoderICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Apr 2024, Seoul, France. pp.7010-7014, ⟨10.1109/ICASSP48485.2024.10446649⟩ |
|
|
Annealed Multiple Choice Learning: Overcoming limitations of Winner-takes-all with annealingNeurIPS 2024 : 38th Conference on Neural Information Processing Systems, Dec 2024, Vancouver, Canada |
|
|
Explainable by-design Audio Segmentation through Non-Negative Matrix Factorization and ProbingInterspeech 2024, International Speech Communication Association (ISCA), Sep 2024, Kos / Greece, France |
|
|
3MAS: a multitask, multilabel, multidataset semi-supervised audio segmentation modelSpeaker and Language Recognition Workshop - Odyssey, Jun 2024, Québec (CA), Canada |
|
|
Multi-microphone Automatic Speech Segmentation in Meetings Based on Circular Harmonics FeaturesInterspeech 2023, international Speech Communication Association (ISCA), Aug 2023, Dublin, Ireland |
|
|
Traitement multi-microphone pour la segmentation automatique de la parole en réunion16ème Congrès Français d'Acoustique, Apr 2022, Marseille, France |
|
|
Microphone Array Channel Combination Algorithms for Overlapped Speech DetectionInterspeech 2022 Human and Humanizing Speech Technology, Sep 2022, Incheon, South Korea |
|
|
Détection de parole superposée distante à l’aide d’une antenne de microphones34e Journées d'Étude sur la Parole, Jun 2022, Île de Noirmoutier, France |
|
|
Joint speech and overlap detection: a benchmark over multiple audio setup and speech domainsLe Mans Université. 2023 |
|
|
Traitement automatique de la parole en réunion par dissémination de capteursAcoustique [physics.class-ph]. Le Mans Université, 2024. Français. ⟨NNT : 2024LEMA1001⟩ |