Researchers have released Cephalonauts One, a comprehensive fMRI dataset designed for decoding naturalistic speech in the human brain. The dataset, collected from three subjects listening to audio podcasts, offers 30 hours of whole-brain 3 Tesla fMRI data per individual. This release also introduces a benchmark for brain decoding, framed as audio segment retrieval, where the goal is to identify the correct podcast segment corresponding to fMRI activity. Performance on this task has been shown to improve with increased training data. AI
IMPACT Enables new research into brain-computer interfaces and AI models for understanding neural representations of speech.
RANK_REASON The item is a research paper describing a new dataset and benchmark. [lever_c_demoted from research: ic=1 ai=1.0]
- 30 Hours
- 3-Tesla MRI Response to TACE in HCC (Liver Cancer)
- alphaXiv
- Audio podcasts in practical courses in biochemistry - cost-efficient e-learning in a well-proven format from radio broadcasting
- audio segment retrieval
- Brain Decoding
- CatalyzeX
- Cephalonauts One
- CORE Recommender
- DagsHub
- functional magnetic resonance imaging
- Gotit.pub
- Hugging Face
- IArxiv Recommender
- Influence Flower
- ScienceCast
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →