LuxVoice
Giving Luxembourgish a stronger digital voice
Advanced Luxembourgish speech technology spanning natural and expressive TTS, speech resources, multilingual transfer, and human-centered evaluation.
View LuxVoiceBuilding AI that listens to every voice.
Dr. Nina Hosseini-Kivanani
I develop intelligent systems that understand how people speak, express meaning, and communicate across languages.
My work brings together speech technology, natural language processing, multimodal AI, and human-centered computing, with a particular focus on Luxembourgish and other languages underrepresented in today’s AI systems.
Language is more than text. A voice carries identity, emotion, rhythm, culture, and belonging. My research explores how AI can understand this complexity without reducing human communication to words or benchmark scores alone.
The future of AI should not speak with only one voice.
Featured projects
Featured work
VoiceTech4Teens
Helping the next generation shape AI
Four hands-on lab days introduce girls aged 14–18 to voice technology, multilingual AI, speech data, chatbots, and creative prototyping.
Register for 2026 Visit VoiceTech4Teens
About Nina
Research grounded in language, people, and public value.
Dr. Nina Hosseini-Kivanani is a computer scientist working across speech and language technology, multimodal communication, and human-centered computing. She holds a PhD in Computer Science from the University of Luxembourg and leads research and development activities in Luxembourgish speech technology through LuxVoice.
Her work spans multilingual speech and language AI for Luxembourgish and other low-resource settings; multimodal systems that interpret voice, prosody, gesture, and context in ways people can understand; and responsible AI for neurocognitive screening, speech analysis, and human assessment.
Alongside research, she is committed to education, mentorship, scientific outreach, and expanding access to AI for young people and underrepresented groups.
Read the full biography Explore projectsTechnology should learn to accommodate people.
People should not have to change their language, voice, or identity to fit the limitations of technology. I work toward intelligent systems that are more multilingual, expressive, interpretable, and inclusive.
Discuss a Collaboration
About Dr. Nina Hosseini-Kivanani
Language. Voice. Intelligence. People.
My work explores how artificial intelligence can understand human communication in all its linguistic, vocal, emotional, and multimodal complexity.
Human communication is never limited to words.
We communicate through voice, timing, rhythm, pronunciation, gesture, facial expression, cultural knowledge, and context. Meaning emerges from the interaction of these signals rather than from any single channel.
My research investigates how intelligent systems can learn from this complexity while remaining useful, interpretable, and accountable to the people who use them. Multilingual and low-resource environments are not marginal technical problems; they are essential settings for developing more inclusive and capable AI.
I studied experimental and clinical linguistics, speech processing, and language and communication technologies before completing my PhD in Computer Science at the University of Luxembourg. This background shaped a research approach that connects speech, language, vision, health, and human behavior.
My earlier work examined how speech, handwriting, drawing, and machine learning could support the study of Alzheimer’s and Parkinson’s disease. That work introduced a question that still guides my research:
When an AI system makes a prediction, can people understand and trust the reasoning behind it?
Predictive performance alone is not sufficient in sensitive domains. Systems must be interpretable, carefully evaluated, and designed to support human expertise rather than obscure or replace it.
Outside formal research, I enjoy cooking, discovering food cultures, walking, hiking, music, and conversations that bring together people from different backgrounds. Cooking reflects qualities I value in research: experimentation, precision, adaptation, and creativity. Walking and hiking create space for observation, reflection, and new ideas.
Why multilingual AI matters
Smaller-language communities often receive weaker speech recognition, less natural synthetic voices, fewer language tools, and limited access to new AI applications. This affects digital participation, cultural representation, accessibility, education, and public services.
Supporting Luxembourgish through synthesis, recognition, multilingual modeling, corpus development, and human evaluation also means supporting the people and identities represented by the language.
Building AI for people
AI should be introduced only where it provides genuine value. My work emphasizes human-centered evaluation, interpretable outputs, collaboration with domain experts, responsible use of human data, cultural diversity, and preservation of human judgment.
I see AI as a collaborative technological layer: one that extends human capabilities without making human knowledge, responsibility, and experience invisible.
Research beyond the laboratory
I enjoy developing research from concept and data collection through modeling, evaluation, publication, and public communication. My collaborations connect researchers, engineers, linguists, educators, media professionals, schools, institutions, and industry partners.
Education and inclusion
My teaching emphasizes experimentation, critical thinking, and meaningful real-world problems. Through VoiceTech4Teens, young people learn that they can understand, question, improve, and shape AI—not merely use it.
AI should learn to understand human communication in all its richness.
People should not have to adapt their language, voice, culture, or identity to fit the limitations of technology.
Projects
From research questions to datasets, systems, and public impact.
I build speech and language technologies for low-resource and multilingual settings — and turn that research into concrete systems, resources, and outreach.
A capable model is only one stage of a useful system. The work below asks what the system learns from, which languages are represented, how uncertainty is communicated, and whether the technology supports human agency.
Three directions shape the projects below.
Multilingual speech and language AI
Speech synthesis, recognition, language resources, and evaluation for Luxembourgish and other languages with limited data.
Multimodal and human-centered AI
Voice, prosody, gesture, and context — with feedback that remains transparent and useful for people.
AI for health and human assessment
Interpretable methods for neurocognitive screening and biomedical language technology that support, not replace, expert judgment.
Project 01 · AI education and outreach
VoiceTech4Teens
Girls in multilingual speech and language AI lab days
A free four-day program where beginners explore how speech becomes data, test multilingual AI, learn about chatbots and synthetic voices, build a team prototype, and present it in a final demonstration.
The goal is not only to teach young people how to use technology, but to show that they can question, create, and shape it.
Project 02 · Luxembourgish speech technology
LuxVoice
LuxVoice develops high-quality speech technology for Luxembourgish, focusing on text-to-speech synthesis, expressive voice generation, speech-data creation, human evaluation, and robust low-resource modeling.
It addresses limited training data, multilingual influence, dialect variation, and code-switching through multilingual transfer, resource development, adaptation, and native-speaker evaluation.
Main objectives
- Natural Luxembourgish TTS
- Expressive and emotional speech
- High-quality speech resources
- Multilingual transfer
- Listening studies and automatic evaluation
Why it matters
Voice interfaces, accessibility tools, media systems, and educational applications depend on speech technology. LuxVoice helps ensure Luxembourgish can participate in that future.
Project 03 · Knowledge technologies
Knowledge graphs for low-resource languages
From entity mentions to structured knowledge
This international COST Action research examines multilingual entity detection, disambiguation, Wikidata linking, NIL detection, triple generation, validation, and knowledge-graph enrichment for local and multilingual information.
Mention detectionEntity disambiguationMultilingual linkingNIL detectionTriple generationValidationGraph enrichment
Leadership
- KGELL · Management Committee Member · Working Group 1 Vice-Chair
Knowledge Graphs in the Era of Large Language Models (COST Action CA24121) — multilingual entity disambiguation, linking, triple generation, validation, and graph enrichment for low-resource languages (2026 – Present)
Research visits & collaboration
- COST STSM · Istanbul Technical University
Research visit on knowledge graphs for low-resource languages, including multilingual entity linking, local entities, NIL detection, triple generation, and graph enrichment
Hackathons & hands-on events
- Hackathon on Metadata and Provenance for Knowledge Graphs — University of Oviedo, Spain; collaborative work on metadata, provenance, and knowledge-graph quality
Project 04 · Luxembourgish NLP resources
LuxNLP
Language resources and models for Luxembourgish
LuxNLP develops language resources, models, and evaluation frameworks for Luxembourgish natural language processing, with a focus on lexical borrowing, neology, and multilingual modeling.
LuxBorrow
- LuxBorrow: From Pompier to Pompjee, Tracing Borrowing in Luxembourgish
Nina Hosseini-Kivanani, Fred Philippy · Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026), Vol. 11, Issue 16, pp. 3171–3183 · PDF
Neology & LLMs
- Do LLMs Know What Luxembourgish Borrows? Probing Lexical Neology in Low-Resource Multilingual Models
Hosseini-Kivanani, N. (2026) · Proceedings of the Workshop Neology and Large Language Models, pp. 27–38 · European Language Resources Association (ELRA) · DOI
Related Luxembourgish NLP
- Mapping Sentiments: A Journey into Low-Resource Luxembourgish Analysis (ECAI 2024 Workshop LUHME, 2024)
- Does Math Reason the Same Across Languages? Prompting Trade-offs in Multilingual Math Support (2026)
- A Parallel Cross-Lingual Benchmark for Multimodal Idiomaticity Understanding (LREC 2026)
Project 05 · Digital health
View Related Publications
Digital Health AI
Neurocognitive screening and biomedical language technology
This research develops interpretable machine-learning methods for health-related assessment and biomedical knowledge extraction — from handwriting and drawing signals for neurocognitive screening to biomedical term extraction and ontology-enhanced annotation.
PhD thesis
- Machine-learning methods for neurocognitive screening using handwriting and drawing
University of Luxembourg, 2025 — Accepted without revisions; nominated for an Excellent Thesis Award
Principle
Clinical AI must remain decision support. Predictions belong alongside professional expertise, data quality, uncertainty, and the limitations of the experimental setting.
Handwriting & drawing for neurocognitive screening
- Automatic Data Augmentation of CDT, HDT, and PDT Images for Cognitive Screening (2026)
- Alzheimer's Disease Screening with Limited and Noisy On-line Pentagon Drawings (2025)
- Screening of Alzheimer's Disease and MCI through Integrated On-Line and Off-Line House Drawing Tests (2024)
- Ink of Insight: Data Augmentation for Dementia Screening through Handwriting Analysis (2024)
- Blueprint of Tomorrow: Contrasting Off-Line and On-Line Drawing Tasks for Alzheimer's Disease Screening (2024)
- Predicting Alzheimer's Disease and MCI with Off-Line and On-Line House Drawing Tests (2024)
- Better Together: Combining Different Handwriting Input Sources Improves Dementia Screening (2022)
- The Magic Number: Impact of Sample Size for Dementia Screening using Transfer Learning and Data Augmentation of Clock Drawing Test Images (2022)
Biomedical text mining
- Biomedical Linked Annotation Hackathon (BLAH) — Tokyo, Japan; travel grant for biomedical named-entity recognition and linked annotation
- TermHunter: Neural Biomedical Term Extraction and Ontology-Enhanced Definition Generation with Structured Prompting (DETECH 2026)
International collaboration
- eVoiceNet · COST Action
Collaboration across Working Group 1 (Data Standards) and Working Group 2 (Open Resources) (2026 – Present)
Collaborative network
Built with partners across academia, media, and funding.
Let’s build language technology that reflects real human diversity.
I welcome collaboration in multilingual speech, low-resource NLP, expressive speech, multimodal communication, interpretable AI, knowledge graphs, healthcare, and education.
Discuss a Research Collaboration
Publications
Research across speech, language, multimodal communication, and human-centered AI.
2026
LuxEmo: Expressive Text-to-Speech Corpus for Luxembourgish
arXiv preprint arXiv:2606.31947 (2026)
LuxSQA: Ask Me in Luxembourgish with TTS-Augmented Spoken Question Answering
arXiv preprint arXiv:2607.02763 (2026)
PIAS: Transparent Multimodal Assessment of Presentation Delivery
Educational Data Mining (EDM), 2026 — Accepted
Does Math Reason the Same Across Languages? Prompting Trade-offs in Multilingual Math Support
2026
TermHunter: Neural Biomedical Term Extraction and Ontology-Enhanced Definition Generation with Structured Prompting
DETECH 2026 — Accepted
A Parallel Cross-Lingual Benchmark for Multimodal Idiomaticity Understanding
LREC 2026
Automatic Data Augmentation of CDT, HDT, and PDT Images for Cognitive Screening
Lecture Notes in Artificial Intelligence, Springer (2026) — Accepted
PolyFrame at MWE-2026 AdMIRe 2: When Words Are Not Enough: Multimodal Idiom Disambiguation
arXiv preprint arXiv:2602.18652 (2026)
LuxBorrow: From Pompier to Pompjee, Tracing Borrowing in Luxembourgish
Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026), Vol. 11, Issue 16, pp. 3171–3183
Do LLMs Know What Luxembourgish Borrows? Probing Lexical Neology in Low-Resource Multilingual Models
Proceedings of the Workshop Neology and Large Language Models, pp. 27–38. European Language Resources Association (ELRA)
2025
A Hybrid Framework for Neologism Validation using LLMs and Lexical Knowledge Graphs
1st International Workshop on Terminological Neologism Management (NeoTerm 2025), June 18, 2025, Thessaloniki, Greece
Speaker Verification Enhancement via Speaking Rate Dynamics in Persian Speechprints
Proceedings of the 14th International Conference on Pattern Recognition Applications and Methods (ICPRAM), Vol. 1, pp. 665-672, 2025
Efficient Automatic Data Augmentation of CDT Images to Support Cognitive Screening
Proceedings of the 17th International Conference on Agents and Artificial Intelligence (ICAART 2025), Vol. 3, pp. 600-607, 2025
Voices of Luxembourg: Tackling Dialect Diversity in a Low-Resource Setting
Proceedings of the Third Workshop on Resources and Representations for Under-Resourced Languages and Domains (RESOURCEFUL-2025), 2025
The Prosody of Cheering in Sports Events: The Case of Long-Distance Running
Phonetica 82.6 (2025) pp. 489–524. De Gruyter, 2025
Alzheimer's Disease Screening with Limited and Noisy On-line Pentagon Drawings
Proceedings of the International Graphonomics Society (IGS), 2025
2024
Screening of Alzheimer's Disease and Mild Cognitive Impairment through Integrated On-Line and Off-Line House Drawing Tests
Proceedings of the International Conferences on Applied Computing and WWW/Internet 2024
Ink of Insight: Data Augmentation for Dementia Screening through Handwriting Analysis
Proceedings of the 2024 8th International Conference on Medical and Health Informatics, 2024
Blueprint of Tomorrow: Contrasting Off-Line and On-Line Drawing Tasks for Alzheimer's Disease Screening
International Conference on Intelligent Data Engineering and Automated Learning (IDEAL), 2024
Predicting Alzheimer's Disease and Mild Cognitive Impairment with Off-Line and On-Line House Drawing Tests
2024 IEEE 20th International Conference on e-Science (e-Science), 2024
Mapping Sentiments: A Journey into Low-Resource Luxembourgish Analysis
ECAI 2024 Workshop LUHME, 2024
2023
ASRLux: Automatic Speech Recognition for the Low-Resource Language Luxembourgish
ICPhS 2023, Prague, 2023
Better Together: Combining Different Handwriting Input Sources Improves Dementia Screening
2023 IEEE 19th International Conference on e-Science (e-Science), 2023
The Magic Number: Impact of Sample Size for Dementia Screening using Transfer Learning and Data Augmentation of Clock Drawing Test Images
2023 IEEE International Conference on E-health Networking, Application & Services (Healthcom), 2023
User Requirement Analysis for a Real-Time NLP-Based Open Information Retrieval Meeting Assistant
European Conference on Information Retrieval, 2023
LUX-ASR: Building an ASR System for the Luxembourgish Language
2022 IEEE Spoken Language Technology Workshop (SLT), 2023
Exploring the Use of Phonological Features for Parkinson's Disease Detection
20th International Congress of the Phonetic Sciences (ICPhS 2023), 2023
2022
The Prosody of Cheering in Sport Events
Proc. Interspeech 2022, South Korea (2022) pp. 5283–5287
IRRMA: An Image Recommender Robot Meeting Assistant
International Conference on Practical Applications of Agents and Multi-Agent Systems, 2022
For a full list of publications, please refer to my Google Scholar page.
Activities
Research has greater value when it creates connections.
My academic work extends beyond publications through research leadership, teaching, mentoring, scientific outreach, international collaboration, conference organization, and public communication.
Leadership & International Collaboration
I contribute to European and international networks focused on multilingual technologies, language resources, knowledge graphs, speech, responsible AI, and interdisciplinary collaboration.
COST Actions
- KGELL, COST Action — Management Committee Member & Working Group 1 Vice-Chair, Knowledge Graphs in the Era of Large Language Models — multilingual entity disambiguation, linking, triple generation, validation, and graph enrichment for low-resource languages, 2026 – Present
- eVoiceNet, COST Action — Working Group 1 (Data Standards) and Working Group 2 (Open Resources), 2026 – Present
- ENEOLI, COST Action — Management Committee Member, European Network on Lexical Innovation, 2025 – Present
- UniDive, COST Action — Management Committee Member, Universality, Diversity and Idiosyncrasy in Language Technology, 2025 – Present
- GoodBrother, COST Action — Management Committee Member, Network on Privacy-Aware Audio- and Video-Based Applications for Active and Assisted Living, 2022 – September 2024
Opening Access to Artificial Intelligence
WeSTEM+
Volunteer supporting women and girls in science, technology, engineering, and mathematics in Luxembourg · 2024 – Present
Public engagement
Panels, school outreach, media communication, public events, and discussions on Luxembourgish language technology, responsible AI, and digital inclusion.
Talks, Panels & Media
Panelist — AI Café 2025: Luxembourg's AI Strategy: Accelerating Digital Sovereignty 2030?
PanelParticipated as a panelist discussing Luxembourg's national AI strategy and digital sovereignty in the context of low-resource language technology and responsible AI deployment.
Featured in Press — Chambre des Députés & University of Luxembourg: LuxASR Collaboration
MediaFeatured in press coverage of the LuxASR collaboration between the University of Luxembourg and the Chambre des Députés for speech recognition of parliamentary proceedings.
Featured: Women Shaping the Future of AI — HeadStart EU Project
InterviewFeatured as a woman leader in AI research, sharing the journey from linguistics to speech and neurodegenerative disease research.
Teaching & Mentoring
Courses Taught
- Introduction to Image Processing — Digital Learning Hub, March & November 2024
- Introduction to Natural Language Processing — Digital Learning Hub, March & December 2024
- Data Science for Humanities — BSc course, summer semesters 2020–2023
- Database Management 1 — BSc course, winter semester 2023
Student Supervision & Interns
- Mikhail Shustrov — Summer 2026
- Louis Alfred Antoine Colbus — Winter 2025
- Grzegorz Piotrowski — Winter 2025
- Sven Kuffer — Winter 2023
- Julien Kühn — Summer 2023
- Daria Antropova — 2022–2023
- Silvana Belegu, Omar Quardighi, Vladimir Blizniukov — research interns 2022–2023
Trainer — 2nd UniDive Training School, Yerevan
Delivered training on SpeechLLMs in low-resource settings. Topics: tokenization, phoneme alignment, prompt engineering, corpus augmentation for Luxembourgish.
Trainer — GoodBrother COST Action Training School, Skopje
Delivered a training session on data augmentation techniques for small neurodegenerative disease datasets at the 1st GoodBrother Training School.
Organizing
- SPEAKABLE2026 @ LREC2026 · Co-Organizer · 2026
- SPELLL 2026 · Proceedings Editor · 2026
- TEICAI @ EACL 2024 · Co-Organizer · 2024
- GLORIA Hackathon · Project Lead · Tokyo · 2025
- BNAIC/BeneLearn 2021 · Website Manager · 2021
Academic Service
Reviewing
- Interspeech 2026
- IEEE Spoken Language Technology Workshop (SLT) 2026
- UniDive Winter School · 2025
- 2nd International Workshop on Causality, Agents and Large Models (CALM) · 2025
- PrivAAL: 2nd GoodBrother Workshop on Privacy-Aware Assistive Technologies · 2024
- ACM Conversational User Interfaces (CUI) · 2022
- International Graphonomics Society Conference (IGS) · 2021
Experience
An interdisciplinary career connecting AI, language, speech, health, education, and human-centered technology.
Download Full CVLast updated July 2026
Research Experience
Researcher and Project Manager — RTL Lëtzebuerg & University of Luxembourg
Luxembourg | Apr 2025 – Present
Leading research and development within LuxVoice, an FNR-supported initiative for Luxembourgish speech and voice technology.
- Research planning and project coordination
- Speech-corpus and TTS development
- Expressive and emotional speech research
- Multilingual transfer and human listening studies
- Academic–industry collaboration, publication, and outreach
Researcher — LuxASR, University of Luxembourg
Belval, Luxembourg | Oct 2022 – Present
Developing ASR for Luxembourgish, including dialect-aware recognition and model optimization for low-resource speech settings.
Research Visit — SpeechTEK, FBK
Trento, Italy | Sep 2025 – Oct 2025
Worked on TTS for low-resource languages (Maltese and Luxembourgish), including data augmentation using KNN-VC.
Research Visit — University of Genova
Genova, Italy | Jun 2023 – Jul 2023
Developed self-supervised learning pipelines for medical AI, from dataset curation to pretraining and downstream fine-tuning.
Research Assistant — ZAS Berlin
Berlin, Germany | Jan 2020 – Mar 2022
Automated phonetic analyses with PRAAT scripts and studied whispered versus normal speech through acoustic modeling.
Education
PhD in Computer Science
University of Luxembourg, FSTM | Jan 2021 – Apr 2025 | Luxembourg
Thesis: machine-learning methods for neurocognitive screening using handwriting and drawing. Accepted without revisions and nominated for an Excellent Thesis Award.
Supervisors: Prof. Luis A. Leiva and Prof. Christoph Schommer
MSc in International Experimental and Clinical Linguistics (IECL)
University of Potsdam | Apr 2017 – Jan 2021 | Germany
Thesis: Automated Intelligibility and Phonological Analysis in Parkinson's Disease Patients
Awarded DAAD Grant
MSc Erasmus Mundus — Language and Communication Technologies (LCT)
Double-degree program | Sep 2017 – Mar 2020 | Netherlands & Italy
Thesis: A deep learning approach to spoken proficiency scoring for non-native speakers
Awarded Erasmus Mundus Scholarship
Awards & Grants
Excellent Thesis Award Nomination — PhD in Computer Science, University of Luxembourg
2025Best Oral Presentation — "Ink of Insight", ICMHI 2024
2024 · Japan3rd Place — Digital Medical Devices Competition
2024 · Luxembourg / BarcelonaTravel Grant — Biomedical Linked Annotation Hackathon 9
2025 · JapanSTSM Grant — Machine Learning Genoa Center (MaLGa Lab)
2023 · ItalyDAAD Grant — Six-Month Master's Thesis Grant
2020 · GermanyTravel Grant — Lisbon Machine Learning School (LxMLS 2020)
2020 · PortugalTravel Grant — AI Summer School PAISS
2019 · FranceErasmus Mundus Scholarship — MSc in Language & Communication Technologies
2017–2019 · Netherlands & ItalySelected Grants & Fellowships
FNR Industrial Fellowship
Funding: XXLuxVoice — Multilingual speech technology for Luxembourgish
FNR PSP-Classic Grant
Funding: 34,542 EURVoiceTech4Teens — AI education and multilingual speech technology outreach
Biomedical Linked Annotation Hackathon Travel Grant
Funding: 4,000 EURBLAH7, Tokyo, Japan
DAAD Master's Thesis Grant
Funding: 2,400 EURGerman Academic Exchange Service — 400 EUR per month for 6 months
Erasmus Mundus Scholarship
Funding: 43,000 EUR totalEuropean Master's program in Language and Communication Technologies (EM LCT)
News & Updates
Research publications, project developments, events, training, and outreach.
Jul2026
Registration open for VoiceTech4Teens 2026
Registration is open for a free program introducing girls aged 14–18 in Luxembourg to multilingual AI, speech technology, chatbots, and creative prototyping across four Saturday lab days at LIST Belvaux.
Apply now Learn moreJun2026
LuxEmo and LuxSQA preprints on arXiv
Two contributions on Luxembourgish expressive speech and spoken question answering are now available as preprints, including LuxEmo, an expressive TTS corpus and evaluation framework.
View publicationsJun2026
Research visit at Istanbul Technical University
Completed a research visit on knowledge graphs for low-resource languages, addressing multilingual entity linking, local entities, NIL detection, triple generation, and graph enrichment.
May2026
Contributions presented at LREC 2026
Contributed to research on Luxembourgish lexical borrowing, multilingual idiomaticity, and language resources.
View publications2026EDM
PIAS accepted at Educational Data Mining 2026
The work combines interpretable prosodic and gesture-based features to support transparent formative feedback on oral presentation delivery.
2026DETECH
TermHunter accepted at DETECH 2026
TermHunter combines neural biomedical term extraction, ontology information, structured prompting, and definition generation.
Jan
2026
Trainer at 2nd UniDive Training School, Yerevan
Delivered a training session on SpeechLLMs in low-resource settings at the 2nd UniDive Training School in Yerevan, Armenia. Topics included tokenization, prompt engineering, and corpus augmentation for Luxembourgish.
Jan
2026
Paper Accepted — Springer Lecture Notes in Artificial Intelligence (2026)
"Automatic Data Augmentation of CDT, HDT, and PDT Images for Cognitive Screening" accepted for publication in Springer's Lecture Notes in Artificial Intelligence.
Sep
2025
Research Visit at FBK Trento — TTS for Low-Resource Languages
Visited the SpeechTEK group at Fondazione Bruno Kessler (FBK), Italy, working on TTS systems for low-resource languages (Maltese, Luxembourgish) with data augmentation via KNN-VC.
Apr
2025
Started the LuxVoice research position
Joined RTL Lëtzebuerg and the University of Luxembourg to lead research on Luxembourgish speech synthesis and expressive voice technology through LuxVoice.
Apr
2025
Doctoral thesis nominated for an Excellent Thesis Award
The PhD thesis in Computer Science was accepted without revisions and nominated for an Excellent Thesis Award at the University of Luxembourg.
Nov
2024
Best Oral Presentation Award at ICMHI 2024
Received the Best Oral Presentation award for "Ink of Insight: Data Augmentation for Dementia Screening through Deep Learning" at the 8th International Conference on Medical and Health Informatics, Japan.
Nov
2024
Paper Published at IDEAL 2024
"Blueprint of Tomorrow: Contrasting Off-Line and On-Line Drawing Tasks for Alzheimer's Disease Screening" published at the International Conference on Intelligent Data Engineering and Automated Learning.
Sep
2024
Paper Published at IEEE e-Science 2024
"Predicting Alzheimer's Disease and Mild Cognitive Impairment with Off-Line and On-Line House Drawing Tests" published at the IEEE International Conference on e-Science, Osaka, Japan.
Contact
Let’s connect.
I welcome conversations about research collaboration, multilingual technology, invited talks, training, student supervision, scientific outreach, and responsible AI.
Collaboration areas
Topics I would be glad to discuss
Luxembourgish speech & language technologyLow-resource TTS & ASRExpressive speechMultilingual NLPSpeech & LLMsMultimodal communicationInterpretable AIKnowledge graphsAI educationResponsible clinical AI
For collaboration requests: please briefly describe the topic, organisations involved, expected contribution, timeline, and relevant funding or project context.
Typical response time: 2-5 business days.
The most interesting AI questions begin with people.
If you are working on multilingual technology, speech, communication, education, or responsible AI, I would be glad to hear from you.
Send an Email