LibriBrain: MEG Recordings of Brain Activity During Audiobook Listening

LibriBrain contains Magnetoencephalography (MEG) recordings of brain activity synchronized with audiobook stimuli. The dataset was first open-sourced for the 2025 PNPL Competition and includes event annotations and stimulus audio files. It was used to evaluate word decoding from brain data in the paper 'MEG-XL: Data-Efficient Brain-to-Text via Long-Context Pre-Training'.

Use Cases

• Train models for word decoding from brain activity based on MEG recordings.
• Analyze neural responses to auditory stimuli based on synchronized audiobook and MEG data.
• Benchmark brain-computer interface algorithms using the event annotations provided.
• Fine-tune pre-trained models for brain signal analysis as referenced in the MEG-XL paper.

Strengths

• Includes multiple data modalities: MEG recordings, event annotations, and stimulus audio.
• Organized by book, providing a structured context for analysis.
• Was used in published research ('MEG-XL') for evaluating word decoding.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface user 'pnpl'.

Collection Method

MEG recordings collected during audiobook listening.

Freshness

Last updated 2026-04-09 18:48:04; freshness should be verified.