🎓 Today we attended the FBK PhD Day, and our amazing students presented their research lines:
@apierg.bsky.social: gender-inclusive MT
@linaconti.bsky.social: explainable speech translation
@zhihangxie.bsky.social: long-form SpeechLLMs
@uzumakidhairya.bsky.social: resource-efficient translation

Voice gender diversity: expression, perception and acoustics Abstract. The voice is a key marker of gender identity, yet gender non-conforming speakers often diverge from binary vocal norms. This review synthesizes f

💬 Pick of the week @fbk-mt.bsky.social: "Voice gender diversity: expression, perception and acoustics."

A great review of gender-diverse voices. Covers acoustic properties, perception and the gap between self-expression of gender and how voices are heard.

royalsocietypublishing.org/rsos/article...

🎙️ Pick of the week @fbk-mt.bsky.social: "Lost in Transcription, Found in Distribution Shift"

Traditional WER masks dangerous hallucinations in ASR systems. This paper proposes a dedicated metric to detect them.

aclanthology.org/2025.finding...

The Unheard Alternative: Contrastive Explanations for Speech-to-Text Models Contrastive explanations, which indicate why an AI system produced one output (the target) instead of another (the foil), are widely regarded in explainable AI as more informative and interpretable th...

🎉 Excited to share that my paper "The Unheard Alternative" was accepted to @blackboxnlp.bsky.social 2025!
We introduce contrastive explanations for speech-to-text, identifying which audio features ST models use to assign a grammatical gender to the speaker.
📄 Preprint: arxiv.org/abs/2509.265...

Pick of the week @fbk-mt.bsky.social: "I Have No Mouth, and I Must Rhyme" 👄 a dive into how LLaMA develops internal phonetic representations despite never hearing speech.
arxiv.org/pdf/2508.02527
#XAI #LLM #interpretability

Languages in Multilingual Speech Foundation Models Align Both Phonetically and Semantically Cross-lingual alignment in pretrained language models (LMs) has enabled efficient transfer in text-based LMs. Such an alignment has also been observed in speech foundation models. However, it remains ...

Pick of the week @fbk-mt.bsky.social : arxiv.org/abs/2505.19606 by @cislmu.bsky.social @barbaraplank.bsky.social. XAI work on cross-lingual alignment in speech-to-text models that disentangles phonetics and semantics. Plus: their XAI insights yield actionable improvements for low-resource languages.

📢 Come and join our group!
We offer a fully funded 3-year PhD position:

📔 Automatic translation with large multimodal models: iecs.unitn.it/education/ad...

📍Full details for application: iecs.unitn.it/education/ad...

📅 Deadline May 12, 2025

#NLProc #FBK

MuPe Life Stories Dataset: Spontaneous Speech in Brazilian Portuguese with a Case Study Evaluation on ASR Bias against Speakers Groups and Topic Modeling Sidney Evaldo Leal, Arnaldo Candido Junior, Ricardo Marcacini, Edresson Casanova, Odilon Gonçalves, Anderson Silva Soares, Rodrigo Freitas Lima, Lucas Rafael Stefanel Gris, Sandra Aluísio. Proceedings...

My pick of the week @fbk-mt.bsky.social: "MuPe Life Stories Dataset" introduces a spontaneous speech corpus that captures Brazilian diversity—across education levels, regional accents & age groups aclanthology.org/2025.coling-...

While we look forward to a sunny Geneva, why wait to join the conversation?

We’ve created a starter pack for our #GITT2025 friends!
🕵️ Follow researchers working on gender bias in MT
💬 Stay up to date and dive into the discussion!

All info at sites.google.com/tilburgunive...