An sEEG dataset of natural Mandarin narrative comprehension

Here, we introduce a large stereo-electroencephalography (sEEG) dataset of 19 participants with 2,882 electrode contacts while listening to an 80-minute Mandarin Chinese audiobook. Alongside the recordings, we provide features of the speech materials, spanning acoustic metrics, segmentation boundaries, and lexical properties.

We validate the recordings with event-related responses and multivariate temporal response function analyses, and provide detailed tutorials for raw-data preprocessing and the validation analyses. This large-scale dataset serves as a benchmark for studying natural Mandarin Chinese speech comprehension, from fine-grained spatiotemporal neural mechanisms to cross-language comparisons.