LST – Laboratoire d'Informatique de l'Université du Mans

Données pseudo-étiquetées de kurde central vers l’anglais pour la traduction de la parole

Corpus: Central Kurdish to English Pseudo-Labeled Data for Speech Translation (Données pseudo-étiquetées de kurde central vers l’anglais pour la traduction de la parole)Licence: CC BY 4.0 licenseAuthor(s): Mohammad MohammadaminiDescription In this repository, you will find large-scale pseudo-labeled data, including Central Kurdish audio translated into English. This dataset contains 1.7 million samples, equivalent to 3,000 hours […]

Machine Learning models explainability for text and audio classification

LST

Seminar from Norbert Tsopze, Professor at Université Yaounde 1 Date: 13/05/2025 Time: 11h00 Place: IC2, Boardroom Speaker: Norbert Tsopze Machine Learning models explainability for text and audio classification Abstract : The interpretability of the Machine Learning models outcomes to the end user is one of the most important properties which favors […]

Towards a Smarter Homophone Correction Tool: A Case Study in Khmer Writing

LST

Seminar from Seanghort BORN, PhD student at LIUM-TEL Date: 7/04/2025 Time: 11h00 Place: IC2, Boardroom Speaker: Seanghort BORN Towards a Smarter Homophone Correction Tool: A Case Study in Khmer Writing Homophone errors are a common challenge in written communication, affecting both high-resource languages like English and low-resourced languages such as Khmer. […]

Thèse en détection de deepfakes

LST

SAVID – Speaker and Audiovisual Interpretable Deepfake Detection Supervisors: Marie Tahon (director) and Aghilas Sini (co-supervisor) at LIUM and Arnaud Delhay (co-director) and Damien Lolive (co-supervisor) at l’IRISA Hosting teams : LST-LIUM et EXPRESSION-IRISA Location : Le Mans Beginning : October 2025 Contact : aghilas.sini(at)univ-lemans.fr, arnaud.delhay(at)irisa.fr Description : The proliferation of text-to-speech and facial […]

Using large ASR models for training lightweight models in low-resource and computation-limited languages

LST

Seminar from Aran Mohammadamini, Post-doc fellow at LIUM Date: 24/03/2025 Time: 11h00 Place: IC2, Boardroom Speaker: Aran Mohammadamini Using large ASR models for training lightweight models in low-resource and computation-limited languages Low-resource languages often suffer not only from a lack of language resources but also from limited computational resources. Recent multilingual […]

Advances in measuring the interpretability of speaker representation spaces

LST

Seminar from Félix Saget, PhD student at LIUM Date: 10/03/2025 Time: 11h00 Place: IC2, Boardroom Speaker: Félix Saget Advances in measuring the interpretability of speaker representation spaces Features extracted by speaker representation models have proven to be potent and versatile, yielding respectable performance in various speaker-related tasks. However, a human user […]

LT4ALL – Bridging Technology & Linguistic Diversity

LST

Aghilas Sini attended the second edition of LT4ALL, an international conference organized by UNESCO, focusing on language technologies for low-resourced and indigenous languages. Over three intensive days, 200+ participants from ~50 countries—spanning academia, NGOs, industry, and governments—gathered to discuss how technology can support linguistic diversity and improve access to knowledge worldwide. 🔹 Day 1: Showcased […]

Towards interpretable representations for audio and speech processing

LST

Seminar from Théo Mariotte, lecturer at LIUM Date: 24/02/2025 Time: 10h30 Place: IC2, Boardroom Speaker: Théo Mariotte Towards interpretable representations for audio and speech processing This seminar is divided into two main parts. The first part reviews my previous research, while the second explores future research directions. In the first section, […]

Information Extraction and Analysis from News videos

LST

Seminar from Sadok Mansouri, ATER at LIUM Date: 13/01/2025 Time: 10h00 Place: IC2, Boardroom Speaker: Sadok Mansouri Information Extraction and Analysis from News videos Information extraction from videos is an important research topic in content-based video indexing and retrieval. Indeed, the visual text present in news videos typically provides rich semantic […]

The Limits of Speech Systems: Navigating Adversarial and Poisoning Threats with Robust Defenses

LST

Seminare from Thomas Thebaud, researcher at CLSP, JHU Date: 18/12/2024 Time : 13h30 Place: IC2, Boardroom Speaker: Thomas Thebaud, The Limits of Speech Systems: Navigating Adversarial and Poisoning Threats with Robust Defense Summary: The widespread adoption of voice-controlled devices and speech recognition systems underscores the critical need for robust security measures […]