I am a first year PhD student in the Language Technology group at the Department of Computational Linguistics, University of Zurich, working with Srikanth Madikeri and Jannis Vamvas.

My research focuses on expressive speech synthesis and speech technologies for low-resource languages, particularly Romansh.

Research Blog (Preparing)

• EmoPilot: Developed and evaluated a label-free framework that predicts emotion cues from target text for expressive speech synthesis. The predicted cues support reference audio retrieval or direct conditioning of the IndexTTS2 backend.
• Romansh TTS: Building expressive long-form text-to-speech systems for the five Romansh idioms and Rumantsch Grischun.

• Yiyang Chen et al. 2026. EmoPilot: Label-Free Emotion Cue Prediction for Zero-Shot Text-to-Speech. Accepted to SLT 2026.
• Sergio Burdisso et al. 2026. SDialog: A Python Toolkit for End-to-End Agent Building, User Simulation, Dialog Generation, and Evaluation. EACL 2026, System Demonstrations. Paper

• October 2026 — FUEN Forum: Participated in the forum and met researchers working on low-resource and minority languages.
• October 2026 — Digidi Demonstration Day: Participated in the event and met native speakers of Romansh.
• June–July 2025 — JSALT: Participated in the workshop, laying the groundwork for SDialog and EmoPilot.