Adonis Asonitis
Member of Technical Staff — Rime
Member of Technical Staff at Rime. I am also an MSc student in Computer Science at ETH Zurich, majoring in Machine Intelligence, where I work on multimodal AI research at the DISCO group (Prof. Wattenhofer), focused on speech generation, large-scale multilingual audio data, and post-training. My recent work includes WorldSpeech, the largest publicly available human-transcribed multilingual speech corpus (65k hours, 80+ languages). I also maintain a public notebook of technical notes from my research.
Featured work
Dataset · #1 trending
WorldSpeech
Largest publicly available human-transcribed multilingual speech corpus. 65k hours · 80+ languages · 150k+ downloads.
Oral · ICML Workshop
MuSE: Multilingual Speech Editing
Highly multilingual, state-of-the-art speech editing. Seamlessly replace a word in an existing utterance.
Oral · Interspeech 2026
Post-Training Speech Enhancement LMs
Speech enhancement language models post-trained with perceptual rewards, reaching state-of-the-art enhancement.