LibriBrain
收藏资源简介:
LibriBrain数据集由牛津大学工程科学系PNPL部门创建,是一个包含52小时单参与者脑磁图(MEG)记录的大型公开数据集。该数据集为非侵入式脑机接口(BCI)研究提供了“通用尺子”,支持语音检测和音素分类等基础解码任务。数据集包含93个会话,总时长为3,139分钟,包含466,230个单词标记和1,511,732个音素标记。LibriBrain数据集旨在推动非侵入式BCI研究的发展,并解决严重瘫痪等医疗问题。
Created by the PNPL group within the Department of Engineering Science, University of Oxford, the LibriBrain dataset is a large-scale public dataset housing 52 hours of single-participant magnetoencephalography (MEG) recordings. It serves as a universal benchmark for non-invasive brain-computer interface (BCI) research, supporting core decoding tasks such as speech detection and phoneme classification. The dataset comprises 93 recording sessions, with a total duration of 3,139 minutes, including 466,230 word tokens and 1,511,732 phoneme tokens. The LibriBrain dataset aims to advance the development of non-invasive BCI research and address medical challenges such as severe paralysis.




