Europe’s linguistic landscape is shaped by national languages, regional varieties, minority communities and countless everyday speech practices. Many of these forms of expression remain difficult for today’s AI systems to recognise or interpret. Our upcoming ELOQUENCE Webcafé will focus on what it takes to design conversational AI that can better understand and reflect this diversity.
This session will explore how speech technologies can learn from real European language use, not only standard forms, but also accents, dialects and underrepresented varieties that shape communication across the continent.
Meet Our Speakers
As part of the discussion, we are pleased to be joined by two researchers from the SpeechTek Lab at Fondazione Bruno Kessler, whose work directly contributes to the multilingual goals of the ELOQUENCE project:
Seraphina Fong is a PhD student in Information Engineering and Computer Science at the University of Trento, Italy and within the SpeechTek Lab at Fondazione Bruno Kessler (FBK). She holds an MSc in Cognitive Science and Computational Linguistics from the University of Trento and a BA in Linguistics and Psychology from Northeastern University in Boston, MA, USA. Her interdisciplinary background informs her research on speech and language processing and multilinguality, with the aim of developing inclusive speech and language technologies that support a wide range of users and communication contexts.
Lorenzo Concina is a Research Scientist at Fondazione Bruno Kessler (FBK), specializing in Automatic Speech Recognition (ASR) and Audio Signal Processing. His current research focuses on extending Large Language Models (LLMs) to the audio modality, specifically within the context of the European ELOQUENCE and IAMI projects. Previously, Lorenzo served as a Machine Learning Specialist at the World Intellectual Property Organization (WIPO), where he optimized and deployed multilingual Speech-to-Text models used to transcribe conferences for several United Nations agencies. He holds an M.Sc. in Computer Engineering from Politecnico di Milano.
Together, Seraphina and Lorenzo will discuss how speech science and engineering approaches can be combined to support more robust multilingual models, what underrepresented linguistic features are most challenging for current systems and how ongoing research can help bridge gaps between real-world speech and AI capabilities.
- Date: December 18, 2025
- Time: 11 AM CET
- Registration link
Join us as we explore solutions, challenges and the road ahead for multilingual AI in Europe!
