@sarapapi

Research Scientist @Google

 Trento, Trentino-Alto Adige
Joined February 2011
  • Tweets702
  • Following434
  • Followers587
24Media
We are still accepting submissions to Last Translation Benchmark (10 accepted in any language for coauthorship 🤩). Also add your model & beat LTB. We added all the latest models: North Small Translate (@cohere), Mistral Large 4 (@MistralAI), Qwen 3.8 (@Alibaba_Qwen) congrats!
3
9
30
2,764
Machine translation is not solved and it will take a while for it to be done arxiv.org/abs/2609.04173

Last Translation Benchmark

For scientific progress, we need benchmarks that test the limits of state-of-the-art models, and evaluation methods that inform us about failure cases. As models get stronger, standard benchmarks...

arxiv.org
5
44
14
188
48,666
There are many things machine translation still can't do. Help us steer the next direction by contributing hard-to-translate inputs (and be on a cool paper).
2
14
5
35
8,118
Sara Papi retweeted
Marco Gaido@mgaido91Jul 1
🚨 DEADLINE EXTENDED! 🚨 More time to shrink your LLMs! The submission deadline for the Model Compression task at #WMT2026 (#EMNLP2026) is now July 15th. 📅 Test data is out. Show us how small you can go without losing translation quality! 🧠🔧 👉shorturl.at/5zXsp
4
4
317
We are at @EAMTee with a large-scale stealth project called "[redacted] Translation [redacted]". Looking for contributors across all languages to challenge MTs. Talk to @PSchmidtova, @sethjsa, @rajaee_sara and @gsarti_.
1
2
1
13
1,473
Sara Papi retweeted
MT Group at FBK@fbk_mtJun 11
We're excited to share that three papers from our lab have been accepted at #Interspeech2026 !! 🍾 #SpeechTranslation #SpeechAI #NLProc #FBK #Interspeech
2
6
8
261
Sara Papi retweeted
Marco Gaido@mgaido91Apr 15
🚀 New Shared Task: Model Compression for Machine Translation at #WMT2026 (co-located with #EMNLP2026)! 📅 Test data out on June 18th, submissions by July 2nd! Can you shrink an LLM and keep translation quality high? 🧠🔧 👉 www2.statmt.org/wmt26/model-… #NLP #ML #LLM #ModelCompression
2
6
11
1,031
Sara Papi retweeted
arXiv Sound@ArxivSoundMar 19
Amirbek Djanibekov, Luisa Bentivogli, Matteo Negri, Sara Papi, "SimulU: Training-free Policy for Long-form Simultaneous Speech-to-Speech Translation," arxiv.org/abs/2603.16924

SimulU: Training-free Policy for Long-form Simultaneous...

Simultaneous speech-to-speech translation (SimulS2S) is essential for real-time multilingual communication, with increasing integration into meeting and streaming platforms. Despite this, SimulS2S...

arxiv.org
1
4
438
Sara Papi retweeted
Tina ✨
@ScuderTinaMar 16
Chinese Gp: Mario Kart edition 🍄🏎️
65
1,762
165
8,713
633,006
Sara Papi retweeted
Meetween@MeetweenEUMar 4
Replying to @sarapapi
@sarapapi introduces MCIF, a new benchmark evaluating 23 SOTA models, revealing gaps in speech–video integration & long-form processing. 🎥 Watch the recording: youtube.com/watch?v=uQrX3iM4… #Meetween #MultimodalAI #LLMs #AIResearch
1
3
105
Sara Papi retweeted
Marco Gaido@mgaido91Feb 10
Hi all! I am looking for 2 emergency reviewers for #ACL rolling review on speech models. If you can help please get in touch with me and/or spread the word! Thanks!
5
7
990
Sara Papi retweeted
Cohere Labs
@Cohere_LabsJan 22
Our Multimodal group is looking forward to hosting @sarapapi tomorrow, Friday, January 23rd for a presentation on "Towards Crosslingual Evaluation of General-Purpose Instruction-Following Models Across Text, Speech, and Vision." Learn more: cohere.com/events/cohere-lab…
1
8
823
Sara Papi retweeted
Slator
@slatornewsJan 8
👉 slator.ch/ItalianBenchmarkEv… A new community-driven initiative evaluates 🔍 large language models using Italian-native 🇮🇹 tasks, with #AI #translation included among the challenge areas. #xl8 #t9n #LLMs @AILC_NLP @vivi_patti @basilepp @peppeatta @matteozio @daniel_sc4 @luisabentivogli @AriannaBisazza @leodecosmo @AnsaScienza @mgaido91 @ItziarGD @MorenoLaQuatra @frannyezooey @arianna_muti @negri_teo @sarapapi @FBK_Research @apierg @teelinsan @gsarti_ @BeatriceSavoldi @Unibocconi @EUI_EU @expertdotai @GalileoEducNet @HeriotWattUni @Idiap_ch @itnewspt @IussPavia @SapienzaRoma @UniBarcelona

Italian Benchmark Evaluates Large Language Models, Includes AI Translation

A new community-driven initiative evaluates large language models using Italian-native tasks, with AI translation among the challenge areas.

slator.com
5
1
8
812
13. Simulstream: Open-Source Toolkit for Evaluation and Demonstration of Streaming Speech-to-Text Translation Systems 🔑 Keywords: AI-generated summary, Streaming Speech-to-Text Translation, incremental decoding, re-translation methods, web interface 💡 Category: AI Systems and Tools 🌟 Research Objective: - Introduce simulstream, an open-source framework for evaluating and demonstrating streaming speech-to-text translation systems, with a focus on long-form audio processing. 🛠️ Research Methods: - Supports incremental decoding and re-translation methods, providing a unified framework for comparing quality and latency. Incorporates an interactive web interface for system demonstrations. 💬 Research Conclusions: - Simulstream is highlighted as a solution to the limitations of SimulEval, offering features for long-form audio processing and an easy demonstration platform. 👉 Paper link: huggingface.co/papers/2512.1…
1
1
1
148
Simulstream: Open-Source Toolkit for Evaluation and Demonstration of Streaming Speech-to-Text Translation Systems arxiv.org/abs/2512.17648 by @mgaido91 @sarapapi

Simulstream: Open-Source Toolkit for Evaluation and Demonstration...

Streaming Speech-to-Text Translation (StreamST) requires producing translations concurrently with incoming speech under strict latency constraints, demanding models that balance low latency with...

arxiv.org
5
1
12
2,539
Sara Papi, Javier Garcia Gilabert, et al., "Hearing to Translate: The Effectiveness of Speech Modality Integration into LLMs,", arxiv.org/abs/2512.16378

Hearing to Translate: The Effectiveness of Speech Modality...

As Large Language Models (LLMs) expand beyond text, integrating speech as a native modality has given rise to SpeechLLMs, which directly process spoken language and enable speech-to-text...

arxiv.org
1
1
394
Sara Papi retweeted
FIA
@fia30 Nov 2025
Leonardo Fornaroli is your 2025 FIA Formula 2 Champion !! 🏆 After winning the FIA F3 championship last year, he seals the FIA F2 championship in Qatar 🇶🇦 #FIA #F2 #QatarGP
7
158
6
1,311
52,420
Sara Papi retweeted
MT Group at FBK@fbk_mt28 Oct 2025
Our next presentation is by @sarapapi: "How real is your real-time simultaneous speech-to-text translation system?" Look for the answer in her TACL paper: direct.mit.edu/tacl/article/… #lt2025fbk
2
7
212
Sara Papi retweeted
MT Group at FBK@fbk_mt21 Oct 2025
🚀 Join us for the LT@FBK day 2025! Discover cutting-edge research and highlights in speech and language technologies from @FBK_research 📅 October 28, 2025 📍FBK, Trento ℹ️ lt-highlights.fbk.eu/
7
9
584
Sara Papi retweeted
MT Group at FBK@fbk_mt10 Oct 2025
Replying to @mgaido91
@mgaido91 introducing SimulStream, an #OpenSource Tool for Simultaneous #Speech #Translation 🗣️🖥️📝 at the @DI_FBK Demo Day at @FBK_research! The tool is going to be released soon. Stay tuned! 👀
3
8
233
Load more