The Machine Translation Unit is heading to EMNLP 2026 in Budapest, Hungary 🇭🇺, October 24 to 29!
We have plenty of research to share this year, covering topics such as speech translation, explainability, simultaneous translation, and multilingual language models. We’re looking forward to meeting researchers and exchanging ideas!
Here are our contributions to the conference and its co-located events:
📚 EMNLP Main Conference
DOA: Training-Free Decoder-Only Attention Policy for Long-Form Simultaneous Translation with SpeechLLMs
Authored by: Sara Papi, Luisa Bentivogli.
Preprint available at https://arxiv.org/abs/2605.31432
Ouvia: A User-Centered Benchmark for Usability Assessment of Speech Translation Systems
Authored by: Giuseppe Attanasio, Beatrice Savoldi, Daniel Chechelnitsky, Marine Carpuat, Matteo Negri, Marteen Sap, Andre Martins.
Preprint available at https://arxiv.org/abs/2606.06177
🛠️ EMNLP DEMO
Simulstream: Open-source toolkit for evaluation and demonstration of streaming speech-to-text translation systems
Authored by: Marco Gaido, Sara Papi, Mauro Cettolo, Matteo Negri, Luisa Bentivogli.
Preprint available at https://arxiv.org/abs/2512.17648
📖 Computational Linguistics
How to evaluate speech translation with source-aware neural mt metrics
Authored by: Mauro Cettolo, Marco Gaido, Matteo Negri, Sara Papi, Luisa Bentivogli.
Paper available at https://direct.mit.edu/coli/article/doi/10.1162/COLI.a.627/136822
🔬 Transactions of the Association for Computational Linguistics (TACL)
SPES: Spectrogram perturbation for explainable speech-to-text generation
Authored by: Dennis Fucci, Marco Gaido, Beatrice Savoldi, Matteo Negri, Mauro Cettolo, Luisa Bentivogli.
Paper available at https://direct.mit.edu/tacl/article/doi/10.1162/TACL.a.684/137205
Hearing to Translate: The Effectiveness of Speech Modality Integration into LLMs
Authored by: Sara Papi, Javier Garcia Gilabert, Zachary Hopton, Vilém Zouhar, Carlos Escolano, Gerard I. Gállego, Jorge Iranzo-Sánchez, Ahrii Kim, Dominik Macháček, Patricia Schmidtova, Maike Züfle.
Paper available at https://direct.mit.edu/tacl/article/doi/10.1162/TACL.a.787/138660
🌍 WMT 2026
Findings of the WMT 2026 shared task on model compression: No free lunch at extreme compression
Authored by: Thamme Gowda, Marco Gaido, Roman Grundkiewicz, Matteo Negri.
Paper available at: https://www2.statmt.org/wmt26/pdf/2026.wmt-1.52.pdf
Findings of the WMT26 Multilingual Instruction Shared Task: Small Models Are Not Yet Polyglots
Authored by: Patricia Schmidtova, Ekaterina Artemova, Seth Aycock, Niyati Bafna, Shobhit Banga, Manmeet Kaur, Tom Kocmi, Philipp Koehn, Danni Liu, Nam Luu, Sara Papi, Beatrice Savoldi, Mariya Shmatova, Hanuman Sidh, Evfrosiniya Zerminova, Vilem Zouhar, Maike Züfle, Pinzhen Chen.
Paper available at: https://www2.statmt.org/wmt26/pdf/2026.wmt-1.50.pdf
FBK’s Submission to WMT26 Model Compression Task: Improving Inference Speed for Quantized Models
Authored by: Dhairya Suman, Luisa Bentivogli.
Paper available at: https://www2.statmt.org/wmt26/pdf/2026.wmt-1.144.pdf
See you in Budapest! Come say hello and let’s talk research! 👋
