Marcely Zanon Boito

Research Scientist (She/Her)

Contact: first.last-name at naverlabs dot com

Hello! I’m a research scientist in speech processing from Porto Alegre, Brazil, currently at NAVER LABS Europe. I work on multimodal dynamic agents for speech and text understanding, with research interests in self-supervised speech representation, multilingual, and efficient speech processing, multimodal interaction, and computational language documentation.

I earned my PhD in Computer Science from the University Grenoble Alpes (2021), focusing on unsupervised word segmentation for language documentation. I also hold bachelor’s degrees in Computer Science (UFRGS, Brazil) and Information Systems Engineering (ENSIMAG, France), and a Master’s in Artificial Intelligence and the Web (MoSIG, UGA, France).

news

Jul 10, 2026

Participated in the PhD Thesis jury of Ryan Whetten, titled Efficient Self-Supervised Learning for Speech Processing and held in Avignon, France.

Jul 2, 2026

Our team reached again first place in the IWSLT 2026 instruction-following short track, tied to the FBK submission.

Jun 15, 2026

Participated in a Dagstuhl seminar From Speech Translation to Multilingual Communication – New Research Challenges.

Jun 8, 2026

Presented an online seminar at ELLIS-Sardine webseminar series SpeechMapper: Efficient training of speech LLMs.

Dec 18, 2025

Presented an online seminar at FBK-MT Training Speech LLMs: Insights and Lessons Learned.