A latency-slope test separates voice- and cue-locked speech detectors in synthetic SEEG; real-data testing was blocked by dataset download failure

This package gives a tested method to find if a SEEG high-gamma speech detector follows the voice onset or the visual word cue. The method splits trials at the median reaction time. It measures the slope beta between the detector half-rise time and the reaction time. Beta = 1 shows a voice-locked detector. Beta = 0 shows a cue-locked detector. In synthetic sessions with known truth, the method gave beta = 0.95 (95% CI 0.90 to 1.06) for a voice-locked response and beta = -0.005 (95% CI -0.020 to 0.013) for a cue-locked response. The planned input was SingleWordProductionDutch-iBIDS (OSF nrgx6, 10 participants). The OSF download failed in this run. Thus, this package has no result from real brain data. It has the pre-analysis plan, the extraction and analysis code, the synthetic test, a figure and a record of the download failure.

speech brain-computer interface · SEEG · high-gamma · latency slope · synthetic validation · voice onset · cue-locked · production-locked