I am a PhD candidate at the University of Bologna, working in the Language Technologies Lab.
My research focuses on multimodal deep learning, language & speech & audio & music modeling, and explainability, with the broader goal of understanding how AI interprets human expressive signals such as voice, language, and musical performance.
- Multimodal AI for Human Expression (speech × text × music)
- Speech & paralinguistic signal modeling
- Multimodal Argument Mining & political discourse analysis
- AI for clinical assessment & decision support
- Music Information Retrieval & creative AI
- Explainable & perceptually-grounded model interpretation
-
Visiting Researcher · Sony AI Barcelona
Adapted large speech–language models to the music domain for musical version identification and audio-based lyrics representation. -
Visiting Researcher · Mila / Québec AI Institute (SpeechBrain Lab)
Benchmarked explainability techniques for speech-based Parkinson’s and depression detection, and developed high-quality listenable explanations grounded in human perception.
“Multimodal AI for Human Expression Understanding”
Investigating how multimodal signals convey meaning across political, clinical, and musical communication settings.
| Platform | Link |
|---|---|
| 🌐 Website | https://helemanc.github.io |
| https://www.linkedin.com/in/eleonora-mancini/ | |
| 🏛️ UniBo Page | https://www.unibo.it/sitoweb/e.mancini/en |
| ✉️ Email | e.mancini@unibo.it |
Python · PyTorch · SpeechBrain · Hugging Face · NumPy
Large-scale training on HPC clusters (Compute Canada, CINECA Leonardo, UniBo HPC)