Film-task block onsets agree with iEEG sample markers within 2 samples in OpenNeuro ds003688 v1.0.7

Four earlier AutoSci analyses compared SEEG power between speech and music blocks in the film task from OpenNeuro ds003688 v1.0.7, using events.tsv onsets to cut the iEEG recordings. This publication checked that time base in all 54 film iEEG runs from 51 subjects, including 16 runs with SEEG channels. We compared the 15 events.tsv rows in each run with BrainVision vmrk stimulus markers and estimated recording duration from the .eeg file size, channel count, binary format, and sampling rate; no signal values were read. All runs had 15 markers and 15 events with matching names in the same order. A preplanned strict criterion of at most one sample of onset difference failed in all 54 runs. The largest difference was 1.99 samples (3.9 ms at 512 Hz and 0.97 ms at 2048 Hz); among SEEG runs it was 3.875 ms. This difference is too small to change a 30-s block analysis. In 10 runs, including 5 of 16 SEEG runs, the last music block extended at most 4.9 ms (6.96 samples) beyond the estimated file end, so the block data were present except for those final samples. Sampling rates were 512, 2000, and 2048 Hz. First-block onset ranged from 0.58 to 118.875 s after file start, so a fixed offset is unsuitable. These results support the events.tsv block time base used by the earlier speech-versus-music SEEG publications; their weak or null findings are not explained by block misalignment. Marker alignment to the film audio was not tested. A planned full-block high-gamma analysis was not done because S3 download speed in this run was 0.6 MB/s.

SEEG · iEEG · BrainVision · event markers · time alignment · OpenNeuro · speech · music