NESPOLE! System has been developed using two scenarios: the tourism scenario and the first aid medical assistance scenario. During the project life three main data collection have been carried on in order to develop the first and the second showcase. During the first year 191 dialogues have been collected. There are 62 German dialogues recorded, 61 Italian, 37 English and 31 French. Particularly an amount of 6 hours of dialogues for Italian and French, 7 hours for English, 8 hours for German has been recorded. Dialogues were about five predefined tourism scenarios. During the last year two major data collections have been carried on: the first one aimed at expanding the tourism scenario and the second one at addressing the medical domain. For the monolingual data collection five tourism scenarios were developed; 66 dialogues were recorded yielding 994.57 minutes of data: 243.52 minutes comprised in sixteen English dialogues, 246 minutes in sixteen German dialogues, 272.52 minutes in seventeen French dialogues and 232.53 minutes in seventeen Italian dialogues. The data collection on the medical domain involved Italian, English and German languages. A total of 49 dialogues were collected. The recording results in a total of 8 hours 25 minutes of audio files.
More great news! 🎉
Our paper “Echoes of Phonetics: Unveiling Relevant Acoustic Cues for ASR via Feature Attribution” was accepted at #Interspeech2025!
Interested in interpretability for speech models? Preprint coming soon!
✍🏼 @mgaido91, @negri_teo, M.Cettolo, @luisabentivogli
Our pick of the week by @BeatriceSavoldi: "Lost in Translation: Artificial Intelligence and the Demand for Foreign Language Skills" by @pmllanos and @carlbfrey (2025)
#AI #translation #MT
Super interesting preprint on the relation between MT improvements and the demand for foreign language skills #pickoftheweek @fbk_mt 📚https://www.oxfordmartin.ox.ac.uk/publications/lost-in-translation-artificial-intelligence-and-the-demand-for-foreign-language-skills
SHADES: a global dataset to uncover AI bias
Over 50 researchers, 16 languages, thousands of interactions analysed: the international SHADES project investigates how generative language models (LLM) reproduce and amplify cultural stereotypes
◾https://magazine.fbk.eu/en/news/shades-the-new-global-dataset-to-monitor-as-ai-reproduces-and-invents-cultural-stereotypes/
🎉 Excited to share our paper “Different Speech Translation Models Encode and Translate Speaker Gender Differently” was accepted at #ACL2025 (main)!
✍🏼 Big thanks to amazing co-authors: @mgaido91, @negri_teo, @luisabentivogli, @andre_t_martins, @peppeatta!
📄 Preprint out soon!