Three papers co-authored by researchers from the MT Unit were presented at Interspeech 2026, the 27th Annual Conference of the International Speech Communication Association (September 28 – October 1, Sydney) 🇦🇺
The first one is “Do What I Say: A Spoken Prompt Dataset for Instruction-Following,” co-authored by Maike Züfle, Sara Papi, Fabian Retkowski, Szymon Mazurek, Marek Kasztelnik, Alexander Waibel, Luisa Bentivogli, Jan Niehues. The paper is already available: https://arxiv.org/abs/2603.09881
The second is “Cross-Attention is Half Explanation in Speech-to-Text Models,” co-authored by Sara Papi, Dennis Fucci, Marco Gaido, Matteo Negri, Luisa Bentivogli. The paper is available here: https://arxiv.org/abs/2509.18010
The last one is: “Better Late Than Never: Evaluating Latency Metrics for Simultaneous Speech-to-Text Translation”, co-authored by Peter Polak, Sara Papi, Luisa Bentivogli, and Ondrej Bojar. The paper is available here: https://arxiv.org/abs/2509.17349
Bravi e Brave to all authors, and thank you to everyone who stopped by our posters in Sydney!
