2026-10-14T11:22:12Z https://www.scidb.cn/oai

10.57760/sciencedb.010sw 2026-09-04T17:06:36Z

2026-09-04 Magnetoencephalography (MEG) based non-invasive Chinese speech decoding doi:10.57760/sciencedb.010sw en The text-MEG dataset was collected at Tongji Hospital, affiliated with Tongji Medical College, Huazhong University of Science and Technology. The dataset contains magnetoencephalography (MEG) recordings from nine healthy participants while they performed a reading-aloud task involving 48 Mandarin monosyllabic words. Each participant contributed approximately 720 trials (48 words × 15 repetitions). The dataset can be used to reproduce the main experimental results reported in Magnetoencephalography (MEG) Based Non-Invasive Chinese Speech Decoding.Brain signals were recorded using a helium-free MEG system manufactured by Beijing X-Magtech Technologies. The dataset includes MEG recordings and experimental procedure data generated using PsychoPy. The procedure data contain the temporal markers required for data segmentation as well as the corresponding sample labels. Code for data preprocessing and reproducing the experimental results is available at https://github.com/ZhihongJia/MASD. Brain-computer interface; Speech decoding; Magnetoencephalography Zhihong Jia Hongbin Wang Yuanzhong Shen Feng Hu Jiayu An Kai Shu Dongrui Wu RESTRICTED dataset http://www.doi.org/10.1088/1741-2552/ae1ea2 Science Data Bank