wordcab/libribrain-meg-preprocessed

LibriBrain MEG预处理数据集包含单个英语说话者听Sherlock Holmes有声书的预处理磁电图（MEG）记录和音素标签。该数据集通过信号空间分离、带通滤波等预处理步骤优化，以加快机器学习模型训练过程中的数据加载速度。数据集分为训练、验证和测试集，并提供了不同样本平均级别的预分组配置，以适应不同的准确性和速度要求。

The LibriBrain MEG Preprocessed Dataset includes preprocessed magnetoencephalography (MEG) recordings and phoneme labels from a single English speaker listening to Sherlock Holmes audiobooks. The dataset is optimized for fast loading during machine learning model training through preprocessing steps such as Signal Space Separation and bandpass filtering. It is split into training, validation, and test sets, with pre-grouped configurations at different sample averaging levels to accommodate various accuracy and speed requirements.