LibriBrain Competition 2025#
Phase 2 (Phoneme Classification) is now live!
We are now accepting submissions for the second phase of the competition, Phoneme Classification, until September 30th. The second phase is fully independent from the first, so you can still join, even if you missed the first phase!
The LibriBrain Competition presented at NeurIPS 2025 invites AI researchers and students to take part in the exciting challenge of non-invasively decoding language from the brain. The societal impact of functioning brain computer interface systems will be substantial - first for people with speech related disabilities, and eventually as a new paradigm for interaction with computers.
We believe that the way forward for advancing this impactful technology is through standardization and easy accessibility for the broader research community. Our goal is to spark an “ImageNet moment” for BCI, lowering the barrier of entry for researchers to contribute to this exciting new field. To this effect, we’re excited to announce
- Deepest MEG dataset to date: 25–50× deeper than most existing datasets
- Plug-and-play Python package:
pip install pnpl - Hands-on tutorials: Train your first model in under an hour
- Community & competition: Prizes 🎁, leaderboards 🌐, Discord 🗣️
We’re excited to see what you’ll build!
